ABC1 - N. Houssami - Overview of imaging tests for assessment of metastases in breast cancer
1. Advanced Breast Cancer Consensus
Conference
3-5 November 2011, Lisbon, Portugal
Overview of Imaging Tests for Assessment
of Metastases in Breast Cancer
Nehmat Houssami
Screening and Test Evaluation Program (STEP)
School of Public Health, Sydney Medical School
2. Overview of Imaging for Assessment of Metastases in BC
(focus on evidence reviews of imaging accuracy)
A/Prof Houssami: No Conflict of Interest
Acknowledgement: STEP receives program grant funding
from Australia’s National Health & Medical Research Council
Acknowledgement
Dr Meagan Brennan (University of Sydney)
Dr Colleen Costelloe (MD Anderson)
3. Background - Imaging for detection or investigation of metastases
or suspected metastases in breast cancer, applied in various
settings, with different purposes
Initial diagnosis of BC (baseline staging) or at subsequent
local relapse and/or suspected distant relapse (restaging)
Investigation of symptoms (or clinical or ‘other test’
abnormalities) raising suspicion of metastatic spread
Surveillance (distant relapse) in women with history of
early BC
Monitoring response to therapy in established metastatic
disease.
4. Review of recent guidelines (January 2011)
NCCN Clinical Practice Guidelines in Oncology: Breast cancer (USA)
2009, 2011 [Carlson et al. JNCCN 2011]
Clinical early-stage or operable IBC Bone scan (localized symptoms, elevated ALP, or IIIA)
Abdominal +/- pelvis CT or US or MRI (abdominal
Optional additional imaging or as directed symptoms or clinical abnormality, elevated ALP or abnormal
by symptoms for stage I (only if symptoms LFT, or T3, N1, M0
present or abnormal staging tests) or for Chest imaging (symptoms)
stage IIA, stage IIB, and stage IIIA (T3, N1,
M0) PET or PET/CT scanning not indicated in routine staging of
clinical stage I, II, or operable stage III.
Locally advanced IBC Bone scan
Abdominal +/- pelvis CT or US or MRI
Optional additional imaging or as directed
by symptoms or abnormal staging tests, for PET or PET-CT scanning discouraged except where other
stage IIIB (T4, N0-2, M0) or IIIC (any T, staging tests are equivocal or suspicious (biopsy of suspected site
N3, M0) recommended alternative)
Bone scan; x-ray symptomatic bones or if bone scan
Metastatic breast cancer (stage IV) abnormal
Chest imaging
Consider abdominal CT or MRI
PET or PET-CT scanning discouraged except where other
staging tests are equivocal or suspicious (biopsy of suspected site
recommended alternative)
5. Review of recent guidelines (January 2011)
NICE / National Collaborating Centre Assess presence and extent of visceral metastases: combination of
for Cancer (UK) 2009 plain radiography, US, CT, and MRI
Advanced BC Guidance Assess presence and extent of bone metastases: bones of axial
Suspected recurrence / metastatic skeleton using bone windows on CT or MRI or bone scintigraphy; assess
breast cancer proximal limb bones for risk of pathological fracture using bone
scintigraphy and/or plain radiography if evidence of bone metastases
→ initial testing should be based on elsewhere
symptoms; once recurrence is suspected (Qualifying statement: Insufficient evidence to support the choice of one
clinically or diagnosed on testing, staging imaging test over another)
recommended
MRI to assess bony metastases if other imaging is equivocal, or
where more information required (for example, lytic metastases
encroaching on the spinal canal)
PET-CT should only be used if conventional imaging is suspicious
but not diagnostic of metastatic disease
ESMO guidelines (Cardoso et al) Annals Chest imaging (x-ray or CT)
of Oncology 2010 Abdominal US, CT or MRI (to identify visceral disease)
General staging recommendations Bone scintigraphy (confirmation of lesions by x-ray, CT, or MRI)
for locally recurrent or metastatic CT and/or MRI of the CNS (if symptoms)
breast cancer PET or PET-CT may be useful for identifying site of relapse when
traditional imaging methods are equivocal or conflicting.
6. Review of recent guidelines (1/2011): Advanced BC guidance (NICE/UK 2009)
Assess for visceral Assess for bone
metastases metastases
(presence/extent):
plain radiography
Axial skeleton: Proximal limb
Ultrasound CT scan or bones:
CT scan MRI or bone Bone scan
MRI scan and/or
radiography
Qualifying statement: Insufficient
evidence to support choice of one
imaging test over another
Imaging suspicious Imaging is equivocal, or
but not diagnostic of more information
metastases? needed?
NO
Yes Yes
PET/CT Imaging confirms or MRI
excludes metastases
7. Two systematic reviews undertaken to provide
background evidence summary for ABC1
N. Houssami & C. Costelloe. Imaging bone metastases in
breast cancer: Evidence on comparative test accuracy. Annals
of Oncology 2011.
M. Brennan & N. Houssami. Evaluation of the evidence on
staging imaging for detection of asymptomatic distant
metastases in newly diagnosed breast cancer (Breast 2011, in
press)
8. Imaging bone metastases in breast cancer: Evidence on comparative
test accuracy. Annals of Oncology 2011 (Houssami & Costelloe )
systematic search of the literature (Jan 2000 to Feb 2011) studies
reporting comparative accuracy for detection of bone metastases (BM)
Subjects with BC as primary, BM from various primary cancers
ineligible
At minimum reported data on TP/FP detection in same subject cohort
Described a reference standard (RS) that was not entirely composed of
one of the tests under comparison
Data extraction: study characteristics/quality appraisal, & quantitative
data on test accuracy
9. Imaging bone metastases in breast cancer: Evidence on comparative test
accuracy. Annals of Oncology 2011; Houssami & Costelloe. Results
Studies meeting criteria: N= 16 (mostly small studies, 15-119 subjects)
Entry into study: suspected BM (symptoms, clinical findings, abnormal
imaging/tumour markers), or staging/re-staging in suspected or known
local or distant recurrence; generally non-consecutive clinical cohorts
Selection of subjects: had conventional test (often BS) → progressed
to newer test (accuracy estimates may be affected by selection bias).
Other limitations noted in quality appraisal (methodology)
Substantial heterogeneity evident for two factors (limits pooling of data)
Quality of the applied reference standard*
Underlying prevalence of BM across studies (study-specific proportion
with BM): median 34% (R: 10% to 67%), partly reflects extent of selection
10. Imaging bone metastases in breast cancer: Evidence on comparative test accuracy.
Annals of Oncology 2011; Houssami & Costelloe. Results
Substantial heterogeneity evident for two factors (limits pooling of data)
Quality of the applied reference standard (RS):
Generally small % confirmed with tissue diagnosis (0-17%)
Clinical +/- imaging follow-up of variable durations
Multidisciplinary team consensus opinion based on available information
Concordance between tests included in RS
Non-verification of test negatives
Within-study combinations of above
Underlying prevalence of BM across studies median 34% (10% to 67)
Does it matter? Above factors combined increase likelihood of misclassification,
& combined with imprecision of study estimates → may account for some of
observed differences in paired test accuracy (vs true differences in
comparative test accuracy)
11. Imaging bone metastases in breast cancer: Evidence on comparative test accuracy.
Annals of Oncology 2011; Houssami & Costelloe. Results
PET or PET/CT vs bone scan (BS)
Seven studies compared FDG-PET with BS
Sensitivity generally similar for both tests (or slightly higher for PET):
median sensitivity for PET 84% (range 77.7-95.2) vs BS 80% (67.0-93.3)
Specificity similar for both tests or higher for PET: median specificity for
PET 92% (88.2-99.0%) vs BS 82.4% (9.1* -99.0)
*low (outlier) estimate from Yang et al for lesion-based data (should be interpreted
cautiously): using data from 10 studies reporting data on BS specificity compared to any
other imaging (excluding Yang) then median BS specificity is 86% (range 68.0%-100%)
• Integrated PET/CT vs BS: Evidence limited to 2 studies, both indicated
higher accuracy for PET/CT than BS (however one of these only looked at
subjects who had discordant results or inconclusive BS), the other (Fuster) gave
100% sensitivity/specificity for PET/CT!
12. Imaging bone metastases in breast cancer: Evidence on comparative test accuracy.
Annals of Oncology 2011; Houssami & Costelloe. Results
Other Imaging comparisons based on too few studies/limited evidence
CT[thorax, abdomen, pelvis] vs BS: 1 study similar sensitivity for CT (98%)
and BS (100%), but CT had higher specificity (100% vs 68%)
SPECT[single-photon-emission CT]) vs PET: much higher sensitivity for
SPECT (85.0%) than for PET (17.0%)
Caution! lesion-based estimates from 900 bone lesions in 15 subjects,
majority of 163 metastatic lesions occurring in 4 subjects (≥ 30 metastatic
lesions each), analytically misleading
reported SPECT may have altered treatment in 1 patient
• CT vs radiography: 1 study, similar (modest) sensitivity for CT (72%)
and x-ray (66%) with same specificity for both tests (100%)
13. Imaging bone metastases in breast cancer: Evidence on comparative test accuracy.
Annals of Oncology 2011; Houssami & Costelloe. Results
Other Imaging comparisons based on too few studies/limited evidence
MRI[conventional] or whole-body MRI vs BS: 2 studies, showed ~ 10% higher
sensitivity for MRI relative to BS, but findings on specificity were inconsistent
Altehoefer 2001 54 subjects MRI (axial) Sen 98% Sp 100%
BS Sen 87% Sp 100%
Engelhard 2004 22 subjects MRI (whole-body) Sen 92.0% Sp 90.0%
BS Sen 83.0% Sp 80.0%
Altehoefer → each of BS or MRI detected sites of BM (in same or alternate anatomic
region) that were not detected by the other test, in some cases this modified local treatment
series comparing whole-body MRI with BS: MRI detected additional non-osseous
metastases (lung or liver) in 1 subject → directed change in chemotherapy; tumor
destruction of vertebral spine with cord compression in 1 subject identified only with MRI
MRI [whole-body MR with diffusion-weighted imaging] vs PET/CT (20 subjects)
Higher sensitivity and specificity for PET/CT than MRI ... But 100% sen/spec for PET/CT
14. Imaging bone metastases in breast cancer: Evidence on comparative test accuracy.
Annals of Oncology 2011; Houssami & Costelloe. Conclusions
Methodologic limitations (Ref Stand / selection issues)
Some evidence that PET (in fewer studies PET/CT), and limited evidence that CT or
conventional MRI
small increments in accuracy of imaging bones relative to BS, where used for
evaluation of suspected lesions and/or bone symptoms or in staging/re-staging
studies have generally compared imaging tests such as PET, PET/CT, CT, or
MRI as add-on tests with BS, hence the evidence is indicative of role of these bone
imaging tests as complementary to BS in an imaging strategy where BS was likely to
have been the initial or baseline test (highest accuracy reported for PET/CT)
Insufficient evidence to base recommendations regarding SPECT or whole-
body MRI for bone metastases
No definitive evidence supporting that any of the imaging tests
discussed in this review can be a replacement to BS in first-line
imaging for bone lesions or symptoms, or staging/re-staging, in BC
15. Review of PET (or PET/CT) for detection of suspected local and/or distant
recurrence, including direct comparison (PET vs conventional Imaging tests, CIT)
and indirect comparison (only PET, or only CIT). Conventional imaging: bone scan,
MRI, CT, conventional work/up
Suspected on basis: symptoms, clinical suspicion, elevated tumour markers,
restaging, abnormal CIT
No distinction between imaging detection of loco-regional and distant recurrence
Quality appraisal: similar methodologic issues/limitations identified in the review
of bone metastases (selection bias/unclear selection, variable RS, incorporation bias,
PET frequently performed after CIT with variable time interval
16. A systematic review of PET and PET/CT for diagnosis of breast
cancer recurrence (Pennant et al) HTA 2010
PET vs CIT (conventional imaging tests) direct comparisons (10 studies): sensitivity
and specificity of PET significantly higher than CIT (each ~10% higher)
Same analysis (sensitivity analysis) for sen/spec based only on studies in which PET
and CIT performed within 1-month period: similar sensitivity, specificity of PET higher
17. Two systematic reviews undertaken to provide
background evidence summary for ABC1
N. Houssami & C. Costelloe. Imaging bone metastases in
breast cancer: Evidence on comparative test accuracy. Annals
of Oncology 2011.
M. Brennan & N. Houssami. Evaluation of the evidence on
staging imaging for detection of asymptomatic distant
metastases in newly diagnosed breast cancer (Breast 2011, in
press)
18. Evaluation of the evidence on staging imaging for detection of asymptomatic distant
metastases in newly diagnosed breast cancer. Brennan & Houssami; Breast 2011 in press
Systematic review (95- 6/2011): Imaging staging for detection of
asymptomatic DM/ reporting detection for any type of DM (not LR recurrence)
Quality appraisal: various biases (similar to those identified in SR of imaging
bones, selection bias, non-verification of test-negatives, incorporation bias)
22 studies/ 3 groups by type of imaging
Conventional imaging only (BS, abdominal US, CXR, CT) combined
Med sensitivity 78% (33-100) / Med specificity 91.4% (67-98)
FDG-PET (some integrated PET-CT) Med sensitivity 100% (78-100) / Med
specificity 96.5% (82-100)
Conventional imaging vs PET (or PET/CT) 5 studies: PET-based higher sensitivity
than conventional (similar median sensitivity as above), variable specificity
19. Evaluation of the evidence on staging imaging for detection of asymptomatic distant
metastases in newly diagnosed breast cancer. Brennan & Houssami; Breast 2011 in press
Med 2.1%
Mean 2.8%
Med 10.3%
Mean 19.4%
Med 31.8%
Mean 23.9%
20. Evaluation of the evidence on staging imaging for detection of asymptomatic distant metastases in newly
diagnosed breast cancer (Brennan & Houssami). Breast 2011 in press
Conventional imaging studies had large data-sets with stage-specific data allowing reliable within-
study analysis (chi-sq test for trend): significant Increase in DM % with increasing clinical stage,
T-stage, or nodal involvement (P<0.0001) [data for clinical stage shown]
Koizumi
Clinical Stage n total n metastasis DM %
0 486 0 0
I 1212 1 0.08%
II 3120 34 1%
III 673 67 10%
TOTAL 5491 102 1.85%
Barratt
Stage n total n metastasis DM %
0 348 0 0
I 992 0 0
II 1041 12 1.20%
III 224 26 11.60%
TOTAL 2605 38 1.45%
Lee
Stage n total n metastasis DM %
0 0 0 NA
I 586 5 0.80%
II 858 6 0.70%
III 395 18 4.60%
TOTAL 1839 29 1.58%
Dillman
Stage n total n metastasis DM %
0 220 0 0
I 502 1 0.19%
II 367 13 3.50%
III 78 12 15.40%
TOTAL 1167 26 2.20%
Kim
Stage n total n metastasis DM %
0 0 0 0
I 448 1 0.20%
II 838 1 0.12%
III 417 25 6%
TOTAL 1703 27 1.50%
21. Evaluation of the evidence on staging imaging for detection of asymptomatic distant metastases in newly
diagnosed breast cancer (Brennan & Houssami) 2011. CONCLUSIONS
very low % of DM in stage I-II / significantly increased with more advanced
presentations (stage III, inflammatory BC, greater node involvement)
PET (PET/CT) performed better → higher sensitivity for detecting DM (~ 25%
increased relative detection) than studies of conventional imaging only BUT
Conventional imaging only studies (relative to PET & PET/CT studies) had:
much lower underlying prevalence of DM
more likely to have consecutive cohorts/ less selection bias
Much larger studies/precise estimates (allowed more detailed stage-specific analysis)
Unclear whether better sensitivity of PET & PET/CT reflects true imaging
capability, or selection issues/underlying risk, biases (probably combination):
Approximation only using median of increased detection (~25% of average 21% DM
prevalence from all studies that included PET or PET/CT): absolute increase in
detection of DM from PET (PET/CT) accounts for ~ 5.3% (effect of selection)
22. Recommend
? Consider prospective multi-centre (paired) trial of staging consecutive
newly diagnosed clinical stage III with conventional imaging + PET/CT
(independent reporting) with adequate follow-up: measure effect on
treatment, clinical & QoL outcomes
Work together → Improve quality of research in imaging
accuracy studies in MBC
THANK YOU
nehmath@med.usyd.edu.au