1. health and fitness
2. disease
3. cancer

Screening MRI in young women
Does it save lives?
How to decrease false positives
Elizabeth A. Morris MD FACR
Chief, Breast Imaging Service
Professor of Radiology
Memorial Sloan-Kettering Cancer Center
New York City, NY
Breast MRI growth
Problem
work up
10%
follow up
8%
Problem
work up
26%
Hi risk
screen
65%
Hi risk
screen
39%
Extent of
disease
27%
follow up
15%
Extent of
disease
10%
MRI Screening for Women at High Risk of Breast
Cancer
• Guidelines generally recommend screening with MRI in addition to
mammography for women at high risk of breast cancer
American Cancer 2007
National Institute for Clinical Excellence 2006
Aged 20 and carry the faulty TP53 gene

BRCA1 or BRCA2 mutation


A first-degree relative with a BRCA1 or BRCA2
mutation

Aged 30–49 and carry the faulty BRCA1 or
BRCA2 gene

Aged 30–39 and >8% 10-year risk of breast
cancer

Aged 40–49 and >20% 10-year risk of breast
cancer

Aged 40–49 with a previous mammogram
showing a dense breast pattern (>12% 10-year
risk of breast cancer)

Lifetime risk of breast cancer of 20–25%, based
on an accepted risk assessment tool (e.g.
BRCAPRO, Claus model, and Tyrer-Cuzick)

Chest radiation between the ages of 10 and 30

Li-Fraumeni syndrome, Cowden syndrome, or
Bannayan-Riley-Ruvalcaba syndrome, or a firstdegree relative with one of these syndromes
High Risk Screening
Author
# CA detected/
# screened
Sensitivity (%)
MG
MR
US
Kuhl
8.1% (43/529)
33% (14/43)
91% (39/43)
40% (17/43)
Warner
9.3% (22/236)
36% (8/22)
77% (17/22)
33% (7/21)
Podo
7.6% (8/105)
13% (1/8)
100% (8/8)
13% (1/8)
TilanusLinthorst
2.8% (3/109)
0%
100% (3/3)
-----
Morris
3.8% (14/367)
------
100% (14/14)
-----
Kriege
2.4% (45/1909)
40% (18/45)
71% (32/45)
-----
Lehman
1.1% (4/367)
25% (1/4)
100% (4/4)
-----
Leach
5.1% (33/649)
40% (14/35)
77% (27/35)
-----
TOTAL
4.0% (172/4271)
84%
(144/172)
Lehman CD. J MRI 2006; 24:964-70
RCT for breast MRI?
• Not likely in the western countries that
currently recommend screening
• Prospective outcome data not available
• Rely on surrogate markers such as size and
nodal status
• $$ Cost-effective?
• No outcome data on whether there is a
reduction in breast cancer mortality
Why is mortality decreased with screening
mammography?
Early detection
40 – 50 % of cancers < 1 cm
20 - 30% should be DCIS
positive nodes in < 20%
DCIS
What size are the cancers detected
on MRI?
Cancers Invasive
detected cancers (%)
Av. Size
(mm)
Median
size (mm)
% ≤ 1 cm
w/o hx
% ≤ 1 cm
w/ hx
Kuhl
31
24 (77%)
n/a*
n/a*
37.5
n/a
Kriege
50
44 (88%)
n/a*
n/a*
43.2
n/a
Leach
35
29 (83%)
14.6
12
44.8
n/a
Warner
22
16 (73%)
11.9
10
46.1
56.3**
Sardanelli
19
15 (79%)
15.1
15
12.5
33.3**
* Kuhl and Kriege gave only size categories, not values, so averages and medians could not be calculated
** Warner and Sardanelli included women with hx of breast cancer, the others did not
48/108 (44.4%) of invasive cancers are ≤ 1cm in women
without personal history of cancer
55/118 (46.6%) of invasive cancers are ≤ 1cm in all women
Method of Detection
Dutch MRISC Study
• 43% cancers detected only on MRI
– 46% of ca in BRCA 1
– 31% of ca in BRCA 2
– 41%of ca in high risk
– 47% in moderate risk
• 9 mm median, 62% ≤ 1 cm
Rijnsburger et al 2010 JCO
MRI screening ↓ advanced stage breast cancer
in BRCA 1 & 2
n
# ca
DCIS/stage I
Stage II-IV
MRI
445
41 (9.2%)
13.8%
7.2%
No MRI
830
76 (9.2%)
1.9%
(p=.01)
6.6%
(p=.02)
• Hazard ratio for stage II – IV for MRI screened
group 0.3 (95% CI, 0.12-0.72; p= .008)
Warner E et al JCO 2011
BRCA 1 versus BRCA 2
•
•
•
•
•
Younger age at diagnosis
Lower mammographic sensitivity
High proportion of interval cancers
Low proportion of DCIS
Larger tumor size at diagnosis
Rijnsburger et al 2010 JCO
Detectability of BRCA cancer on
mammography
• Reduced due to benign features - “pushing
margins”*
• Radiation sensitivity
• In pooled data where all 3 modalities used
– 3/83 (4%) only detected by mammography
– ?justification for replacing mammography in these
patients with MRI
* Kaas et al Eur Radiol 2004
Do high risk women screened with MRI require
mammography as well? (n=687)
• CBE, mammo, US, quality-assured breast MRI, alone or in
different combinations
• 27 CA - 11 DCIS (41%) ; 16 invasive CA (59%)
– 3/27 (11%) node positive
– No interval CA ; no CA at half-yearly US
• CA yield
– US 6/1000; mammo 5.4/1000 NS difference (NS ↑ if
combined 7.7/1000
– CA yield achieved alone by MRI 14.9/1000 significantly higher
– Not significantly improved by adding mammo (MRI plus
mammo 16/1000 or US (MRI plus ultrasound 14.9/1000
Kuhl C et al. J Clin Oncol 2010
Should we screen only with MRI in
BRCA 1 carriers?
• 93 BRCA 1 associated cancers ’03 – ’13
• Screening sensitivity 90/94 (95.7%)
– Mammography 48/94 (51.1%)
– MRI 88/94 (93.6%)
P < 0.0001
• Interval cancers 4/94 (4.3%) (all TN)
• MRI only detected cancer 42/94 (44.7%)
• FFDM only detected cancer 2/94 (2.1%)
– Grade 3 in 50y & gr 2 in 67y
BRCA 1 carriers Recommendation
• Digital mammography
only added 2% to
detection of cancers – no
benefit < 40y
• Screen only with MRI
until age 40
• 1847 screening rounds, 57 cancers
– 53 screen-detected, 1 interval & 3 incidental at Mx
• 37 (65%) invasive (mean≈1cm / 90% node neg)
• sensitivity of MRI vs mammo 86% vs 19%
• Incident cancers, 97% were Stage 0 or 1.
Outcomes
• Of 28 previously unaffected women diagnosed with
invasive cancer
• 1 BRCA1 mutation carrier died following relapse of 3
cm, node-positive breast cancer diagnosed on first
MRI screen at age 48
• 3 patients died of other causes
• None of 24 survivors has distant recurrence (median
follow-up of 8.4 years since dx)
Passaperuma K et al BJC 2012
Take home from this study
• MRI in BRCA1/2 carriers detects majority of
breast cancers at very early stage.
• Absence of distant recurrences of incident
cancers to date is encouraging.
• Longer follow-up is needed to confirm the
safety of breast surveillance over prophylactic
mastectomy
Passaperuma K et al BJC 2012
Dutch MRI Screening (MRISC) – Family
history without proven BRCA mutation
• 1597 women (8370 women years) ‘99-’07
– 47 breast cancers detected (9DCIS)
• Calculated cost per detected and treated cancer
• Microsimulation screening analysis model
• Screening with MRI costs $123,672 (€93,639) per
detected cancer
Saadatmand S et al. Cost-effectiveness of screening women with familial risk for breast
cancer with magnetic resonance imaging. JNCI 2013
Screening with MRI : survival benefit for
familial risk but expensive
• Screening with annual MRI & mammo 35-50 y
– ↓ breast cancer mortality 25% at $134,932 (€102,164) per
life year gained
• Screening with biennial mammo only 35-50 y
– ↓ breast cancer mortality 17% at $54,665 (€41,390) per
life year gained
• Screening in older women
– Costs per detected and treated cancer decreased
– The % of MR detected cancers increased
Saadatmand S et al. Cost-effectiveness of screening women with familial risk for breast
cancer with magnetic resonance imaging. JNCI 2013
Low interval cancer rate needed
for benefit of screening
• 28,061 MRI ‘05-’10
• MRI alternating with mammo/MRI q 6 months
No
Mean
size
(cm)
Invasive
histology
Method of detection
Cancer between
annual MRI
screen
48
(0.2%)
0.7
18/48 (37%)
8(17%) palp
34 (69%) mammo ca++
6(13%) MRI
Screen detected
cancer
120
(0.4%)
0.8
87/120 (73%)
p
Sung J et al MSKCC data
<0.0001
Comparison of DCIS
Low grade
Int grade
High grade
Cancer between
7(23%)
annual MRI screen
17(57%)
6(20%)
Screen detected
cancer on MRI
17(52%)
12(36%)
3(9%)
Screening MRI preferentially detects invasive cancers
and int/high grade DCIS
Most interval cancers are detected as ca++ on mammo
Sung J et al MSKCC data
MRI screening: is it working?
• Likely
• Surrogate markers appear favorable
• Preliminary outcome data appears to support
in BRCA carriers
• Modeling data in familial risk supportive
• Expensive
MRI Screening recommended for
patients with h/o Chest Irradiation
Screening MRI detected additional CA in 4%
– Total of 10 cancers in 9/91 patients
• 4 MRI detected cancers:
– 3 IDC (0.5-1.6 cm), 1 DCIS
• 3 MRI and MG detected cancers:
– 1 IDC (0.8 cm), 2 DCIS
• 3 MRI occult cancers: DCIS
Sung et al. Radiology 2011; 259:65-71
• breast MRI not more sensitive than mammo
for cancer detection
• BUT screening modalities complemented each
other
How to decrease false positives in young
patients on breast MRI
No two breasts look alike
Background parenchymal enhancement & intensities vary
MRI shows us that not all dense
breasts are the same
Variable amounts of enhancement
In order to differentiate from abnormal enhancement
Called Background Parenchymal Enhancement (BPE)
Minimal
Mild
Moderate
Marked
MRI BI-RADS® Lexicon – 2nd edition
Pathologically what are all those dots?
Stippled is a retired term that described a pattern of BPE
Uematsu et al. Eur J Radiology 2011
Radiology
2011
Strong association with enhancement on MRI and breast cancer risk
Risk for Breast cancer increases with
increasing BPE
39 MRI screen detected cancers in 1275 women
Reader 1
Reader 2
Increasing BPE Trend
p < 0.001
p = 0.002
Min/Mild vs
Mod/Marked
OR 10.1 [2.9-35.3]
OR 3.3 [1.3-8.3]
Risk relationship is at least as strong as breast
density
Likely is higher
Patterns of BPE are unique to the
individual
250 initial screening MRI in high risk patients
25% Minimal/34% Mild/24% Moderate/17% Marked
Background parenchymal enhancement had
NO impact on
1. biopsy recommendation rate
2. PPV of biopsy
3. cancer detection rate
Follow up BI-RADS 3 category
• 109/250 (44%) overall
• Malignancy rate of actual lesions followed
• 1/250 (<1%)
• Significant higher call back for
marked/moderate
– Mild 27%
– Moderate 45%
– Marked 58%
BPE should NOT be followed or biopsied
History of LCIS
7 mm invasive lobular cancer
Negative nodes
Cancer detection and call back related to
BPE
• No overall difference in diagnostic accuracy
depending on BPE
• Slightly higher work up of women with higher
BPE
• Perhaps the limitation of BPE on MRI is not as
strong as the impact that density has on
mammographic accuracy
Liu F et al. Wash U Seattle RSNA 2008
Paradoxically cancer detection highest in
patients with highest BPE
•
•
•
•
Minimal
Mild
Moderate
Marked
3.2%
2.4%
5%
5%
?Possibly related to the increased risk
of developing a cancer with BPE
Hambly NM et al AJR 2011
Pre & post Lactation
pre
pre
post
post
Cancer risk elevated while pregnant & first year post partum
Native breast tissue responsive to
hormonal changes
Screening annual exams
HRT started year 3
Efficacy of
treatment
pre
pre
post
post
BPE increases back to pre treatment level
once Tamoxifen stopped
On Tamoxifen
December 2008
Off Tamoxifen
December 2009
Similar effects
seen with AI
pre
post
Ovarian ablation
Pre
Courtesy of Elissa Price MD
Post
Whole breast Radiation Therapy
BPE irradiated breast
Permanent effect
What lesions should I consider following?
First question to ask:
Is it a unique finding or is it part of BPE?
Foci are UNIQUE and stand out from
BPE but otherwise have no suspicious
features
A focus that is unique without clear
suspicious features
Follow up of a unique focus without suspicious features
on an initial screening examination is recommended
Foci that are < 5mm can be small cancers
UNIQUE and have other characteristics that are
suspicious
Foci that are unique and have suspicious features should be biop
Focus that is bright on bright fluid imaging
Post gad T1
MIP
T2
oci that are bright (cyst bright) on T2W or STIR imaging are benign
Coded as BI-RADS 2
Bright on T2W imaging – compare with
vessel or cyst
Benign Lymph Node
Benign Lymph Node
First MRI screen following
lumpectomy, ACT, RT
No pre-op study
Unique Focus with
suspicious features &
not bright on T2W
Lumpecto
site
BI-RADS® 3 for ? Atypical reactive lymph node
At 6 months…..
Beware of a focus that is not bright on T2W or STIR
Mucinous cancer will not be mistaken for a
bright T2W focus
Mucinous cancers have many other suspicious features
other than the high T2W signal
Mucinous cancers – bright on T2
Stable foci over 3 years
A new focus that washes out
Courtesy of Connie Lehman MD
High risk screen – new focus
Targeted US negative
MR biopsy
IDC
New focus – iso to bright on T2
MIP 2010
T2
MIP 2011
postgad
sub
MR biopsy
MR biopsy - severe ADH bordering on DCIS and papillary lesion
Conclusions
• Background parenchymal enhancement is
related to NORMAL hormonal state and
treatment (BI-RADS 1)
• Focus (<5mm) with benign morphology and
signal characteristics is benign (BI-RADS 2)
• Focus that is UNIQUE without complete benign
morphology and signal characteristics may
follow on initial examination (BI-RADS 3)
• Foci that are UNIQUE with suspicious
morphology should be biopsied (BI-RADS 4)
THANK YOU!
Hospital MSKCC
Evelyn Lauder Outpatient
Breast Center MSKCC
Breast MR Imaging Screening in Women with a
History of Breast Conservation Therapy
• Single-screening MR imaging depicted 18.1
additional cancers per 1000 women with a
history of BCT
• Multivariate analysis
– independent factors associated with women with
MR-detected cancers
• younger than 50 years at initial diagnosis (P < .001)
• > 24-month interval between initial surgery and
screening MR imaging (P = .011)
Gweon HM et at Radiology 2014
Breast MRI
• No data exists evaluating screening MRI in dense
breasts
– At least 11 prospective screening MRI trials, however, all in
high risk women, which likely overestimates cancer
detection compared to our population of interest of dense
breasts only
• There is an accruing study : the DENSE trial (Dutch) evaluating
screening MRI in women 50-75 with extremely dense breasts
and negative MG
FaMRIsc Trial
• Breast density as a factor to screen women with familial risk
of breast cancer with MRI – personalized screening
• MC RCT parallel design
– 1: mammo & CBE q y
– 2: MRI & CBE q y ; mammo q 2 y
• Eligibility: 30-55 y ≥ 20% LTR (BRCA excluded)
• Endpoints:
– 1⁰ - # and stage of cancer
– 2 ⁰ - false positives/PPV/sensitivity
Saadatmand S et al. BMC Cancer 2012
Risk of recurrence & contralat cancer
higher in BRCA (n=54) in BCS & RT
• 162 matched sporadic cancer controls for age,
tumor size, and time of surgery - same treatment
• 10 yr IBTR
– 27% for mutation carriers
– 4% for sporadic controls (hazard ratio 3.9; 95%
confidence interval 1.1-13.8; P = 0.03)
• 10 yr CBC
– 25% for mutation carriers
– 1% for sporadic controls (P = 0.03)
Interval breast cancer
11/14/2005 BIRADS 2 MRI
7/7/2006 BIRADS 6 MRI
Mitch Schnall MD PhD
Should we screen more frequently
in BRCA patients?
5 month interval –
Unresolved issue – screening interval
History of LCIS
1 year before
6 months before
7 mm invasive lobular c
Negative nodes
EVA Trial
• Prospective, multicenter study
• Respective contribution of clinical exam,
MG, US, and MRI
• 687 women with ≥20% lifetime risk
Kuhl C et al. JCO 2010; 28:1450-57
Cancer Yield of the Different Imaging Methods
Kuhl C et al. JCO 2010; 28:1450-57
EVA Trial
• CA yield of MRI alone significantly higher
– Slightly improved with MG
– No improvement with US
• PPV:
– MRI: 48%, MG: 39%, US: 36%
• MRI superior for both invasive CA and DCIS
Supplemental Imaging
• The impact of supplemental imaging can be expected
to
–
–
–
–
Reduce mortality
Reduce node positive cancers
Reduce stage II-IV disease
Reduce the interval cancer rate
Key Points
• Women with dense breasts make up a significant
proportion of the screening population
• Studies have shown that women with dense breasts
may have a 1.8-6.0 higher risk of developing breast
cancer
– Density determination and reproducibility remains
varied
• Dense breast parenchyma is also known to mask
pathology in the breast
• Supplemental screening will find more cancers
Digital Breast Tomosynthesis
• 4 large clinical studies (>50 000 women)
(Skaane 2013, Ciatto 2013, Rose 2013, Haas 2013)
• Ciatto (STORM trial): Incremental cancer detection rate with dense
breasts was 2.5/1000
• Skaane: Increased cancer detection was 40% and was similar in
non-dense and dense breasts
• Haas: Recall rate decreased the most in dense breasts.
– For heterogenously dense: 45% reduction
– For extremely dense: 60% reduction
– Confirms that DBT helps reduce the masking effect
Key Points
• Contrast imaging has the highest detection of
additional cancers in high risk women
• CEDM appears to have fewer false positives
compared to MRI without decreased sensitivity
• Mammography is not a perfect test however it is the
only one demonstrated to reduce breast cancer
mortality
What about screening US?
4,897 women
Dense breasts
31 cancers
3/1000 (0.3%) cancer detection rate
Kolb et al. 2002
Screening US – Italian study (n=6449)
• Dense breasts & negative mammogram
• 29/6449 (0.45%) cancers detected by US
• US caused additional work-up in 5% & benign
biopsy in 1%
• Incremental cancer detection higher in <50
than >50y
Corsetti V et al. Eur J Cancer 2008;44:539-44
US Breast Cancer Incidence Rates By
Stage of Disease
60
Cases per 100, 000
50
40
30
20
10
0
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
Year
In Situ Incidence Rates
Stage I Incidence Rates
Stage IV Incidence Rates
Source: SEER
US BREAST CANCER MORTALITY 1975-2007
~30% reduction
% CHANGE IN US FEMALE BREAST CANCER
MORTALITY BY DECADE
1990s
-2.1
1980s
0.5
1970s
-0.2
0.2
1960s
-0.1
1950s
0
1940s
0.8
1930s
-3
-2
-1
0
1
source: ACS, 2003
‘Incidence of Breast Cancer in USA 1975-2008
Increasing
diagnosis of
in situ cancer
with
screening for
women of all
ages
Source: SEER
Breast
cancer
5 year
survival by
stage at
diagnosis
99%
84%
23%
Mortality Reduction in Prospective
Randomized Trials of Mammography
35
HIP
40-64
24 (7-38)
30
Malmo
45-69
19 (-8-39)
25
Two county
40-74
32 (20-41)
20
Edinburgh
45-64
21 (-2-40)
15
Gothenburg
39-59
16 (-39-49)
10
NBSS-1
40-49
-3 (-26-27)
5
NBSS-2
50-59
-2 (-33-22)
Stockholm
40-64
26 (-10-50)
All
39-74
24 (18-30)
0
-5
Swedish 2-county trial:
Screening Impact on Breast Cancer Mortality with 29
Year Followup
Tabar. Radiology 2011; 260:658-663
Relative risk of breast cancer death for those
invited to screen vs not invited = 0.69-0.71
Screening 300 women
for 10 years prevents
one breast cancer
death.
Longer followup (at
least 20 years) shows
©2011 by Radiological Society of North America