1. Evaluation of Tumor Response to Neoadjuvant Chemotherapy
for Breast Cancer
Focus on Breast MRI
After NAC
Jane Wang, MD, PhD
Department of Radiology
Taipei Veterans General Hospital

2. Summary MRI: more accurate than Mam, US MRI techniques
Tumor pattern on MRI after neoadjuvant chemotherapy (NAC)
Measurement of residual disease (RD)
1D, 2D, 3D and correlation with pathology
Factors influence MRI accuracy for size measurement
MRI evaluation of pCR (pathologic complete response)

3. Rationale NAC: downstage the tumor; increase the possibility of BCS
Tumor-free margin- lower long term recurrence (2-7% in 5 yr; vs 22% with positive margin)
Imaging: aid in surgical planning
Pitfalls in evaluation, factors affecting MRI accuracy over, underestimation

4. Sahoo, et al
Arch Pathol Lab Med
2009
Best

5. MRI techniques

6. Technical aspects Most studies use 1.5 T MR scanners DCE MRI Adjunct
1.5T and 3T: no difference
Some studies: 1.5 T superior to 3T?? But insignificant (different readers)
DCE MRI
Adjunct
Proton MRS
DWI/ADC

7. The tumor pattern on MRI after NAC

8. Tumor pattern changes Thibault 2004 Kim 2007 Kim 2012
Pre-C/T: circumscribed nodules vs spiculated tumor; uni- vs multifocal
Post-C/T: shrinkage vs fragmentation
nodules—shrinkage; spiculated– fragment
Kim 2007
Shrinkage vs nested/rim vs mixed
Kim 2012
Pre-C/T: solitary, grouped, separated, replaced
Post-C/T: I, II concentric shrinkage (II with satellite); III, multinodular; IV: diffuse contrast E
I, III good responders; IV nonreponder
Summary: shrinkage vs fragmentation/scattered foci (prone to underestimate)

9. 1. 60 y/o. TNBC
Pre-chemo mammo

10. Pre-chemotherapy

11. Scattered invasive tumor foci in a tumor bed spanning about 2.4*1.1cm
Post-chemotherapy
Scattered invasive tumor foci in a tumor bed spanning about 2.4*1.1cm

12. 2. 35 y/o pre chemotherapy mammo/TNBC

13. Pre-chemotherapy MR

14. Minimal scattered invasive carcinoma cell <1% cellularity
Post-chemotherapy
Minimal scattered invasive carcinoma cell <1% cellularity

15. Threshold criteria of no RD enhancement
NO enhancement at all (spe 54%)(exclude benign inflammation)
Enhancement at tumor bed <= adjacent normal glandular tissue (spe 83%)
Compare with pre-C/T MR

16. Measurement of RD and
Response Evaluation by MRI

17. PERCIST 1.0: PET response criteria in solid tumors
JNCI 2000; 92:
Bethesda Handbook of Clinical Oncology, Appendix IV.
RECIST: Response evaluation criteria in solid tumors. NC: no change.
PERCIST 1.0: PET response criteria in solid tumors
descriptions of methods for controlling the quality of FDG PET
comparability of PET from different time points
quantitative expression of the PET parameter changes
 assessment of overall response
Hyun, et al2016 Radiology

18. WHO
RECIST
Continued

19. WHO
RECIST
Continued

20. MRI measurement of RD
Sensitivity: %; Spe: %; acc: %.
Early DCE MRI phase; occasional late phase
TIC: may change (to less washin, sometimes  persistent)
1D (RECIST; more compatible with patho measure): correlation with patho
2D (WHO; bi-product): response evaluation
3D: response evaluation

21. Correlation of MR and patho
Size (r):
MRI ( )> CBE ( ); US ( )
MR mass (0.85); NME (0.69)
Agreement of response:
MRI ( %) > CBE (19%); US (35%); mammo (26%)

22. MRI kinetics
volume transfer constant (Ktrans), exchange rate constant (Kep), early contrast uptake (ECU) –early prediction response
AUC of Ktrans (0.93) > bi-dimensional size (0.68) (Ah-See 2008 Clin Ca Res)
Sen= predict responders; Spe= predict “non-Res”
ECU Sen 100%, Spe 71% for pCR prediction
These usually for early stage MRI—prediction of final response (Marinovich 2012 Breast)

23. Why MRI over ?
Over:
MRI may include DCIS (but RD only measure invasive ca)
Inflammation around the tumor can enhance
Radiologists measure total extent but pathologists measure only the largest one
Taxanes increase permeability/ contrast uptake
Change of TIC pattern, measure at late phase

24. Why MRI under? Antivascular agent: decreased contrast uptake
Fragmented, scattered tumor foci

25. factors affecting MRI accuracy to measure RD
MRI pattern:
Non-focal (NME): lower correlation than mass
Molecular subtypes, C/T, targeting Tx, age…
TNBC most accurate (r= 0.781)
Usually mass type, shrinkage pattern
ER-pos (or ER pos/ HER neg to 0.504): least accurate (scattered foci)
HER2 (+) acc > HER2 (-)
Herceptin kills scattered foci so increase acc?
(other showed contradictory result)
Age: <45 y/o , higher discrepancy

26. Factors affecting MR acc
Proliferation index Ki-67
More accurate for high Ki-67
more aggressive tumor, more accurate
Taxane: reduce accuracy (scattered foci?)
r=0.5 compared with FEC group (0.89)
Avastin: NS effect on MR accu.
sen for RD detection, 76% vs 70% for with vs without group
Patho: ILC underestimate by MR- NME, multicentric, less solid than IDC
5FU
Epi
cyclo

27. Factors affecting the residual tumor size dx by MR Chen, et al
Factors affecting the residual tumor size dx by MR Chen, et al JSO
N= 98. MRI dx of pCR ((-) at OP)- overall FN 16/98 (16.3%); acc 77/98 (78.6%)
FN significantly lower in:
Tumor type: IDC (P 0.007)
MR scanner: 1.5T (P 0.005)
NS in morphology (mass vs NME), grade (high/ low + intermed); HR, HER2; Regimen
Multivariate:
tumor type, morphology, molecular subtype-predictor for MR-patho size discrepancy
pCR = 44; MR dx 39

28. Chen et al
JSO 2014
MRI-patho
Size discrepancy
AC: doxorubicin
cyclophosphamide

29. Author
/published year
Recruitment period n
Molecular
subtypes r
Acc
*Size discrepancy, cm, mean
(SD)
Moon, 2009 [30]
2006/Jan
-2008/2
195
Overall
TNBC
Lum A
Lum B
HER2(+)
Validation (n=63)
Age>45, HER2(-)
Age>45, HER2(+)/age<45, HER2(-)
Age<45, HER2(+)
0.781
0.504
0.502
0.384
0.849
0.733
0.513
#Moon,
2013 [29]
2006/1
-2010/5
463
Overall: US MR
Lum B/no †tra
HER2(+)/no tra
HER2(+)/with tra
0.552
0.644
0.531
0.583
0.740
0.754
0.608
0.746
0.688
1.39cm(1.44)
Chen, 2011 [35]
2006/11
-2010/10 50
1. HER2(+)
HER2(-)
2. HR(-)/HER2(-)
HR(+)/HER2(-)
3. Ki-67>=40%
Ki-67<=10%
0.5 (0.9)
2.5(3.5) (P=0.009)
1 (1.1)
3(4) (P=0.04)
0.8 (1.1)
3.9 (5.1)
(P=0.06)
Kuzucan,
2012 [24]
2002/5
-2010/2 54
All HER2 (-)
HR(+)
HR(-)
Ki-67>=40%
Ki-67<40%
Sen 76;
spe 100;
acc 83
Sen 88,
spe 88,
acc 88
1.6 (2.8)
0.6 (0.9)
(P=0.05 for mean; P < for range)
0.4 (0.8)
1.2(2.0)
(P=0.05 for mean; P=0.002 for range)

30. Pathologic complete response

31. pCR Definition Typical definition for pCR:
absence of invasive CA (in situ may be present)
Near-pCR: minimal residual invasive cancer
Some studies: neither invasive nor DCIS
pCR rates: %, median 16%.
MR accu for pCR varied with definition
Higher when DCIS excluded, lowest for standard def (detecting RD as true positive)

32. Other Factors– pCR Rates
Higher in younger (<40), larger (>4cm), higher grade (gr 3), higher Ki-67, ER neg, basal-like, HER2 pos.
Lower in lobular ca, lower grade, lower Ki-67, ER pos, Lum A
Higher with taxane based C/T than anthracycline based, Herceptin for HER2 pos.

33. MRI Acc- pCR / Molecular Subtypes
HER2 pos > HER2 neg
HER (-): more likely scattered foci
Herceptin for HER2 (+): clear peripheral foci
Antivascular (Avastin) for Her2 (-): increased scattered foci
Best accu for pCR in TNBC cancer: sen/spe/acc 60, 100, 82%;
PPV in TNBC (100%)> non-lum/HER2 pos 92% > Lum/Her2 pos (45.5%)/ lum/HER2 neg (33.3%) Hayashi, et al. Oncol Lett 2013

34. Conclusion MRI for RD: more accurate than Mam, US
Over and underestimation still occur
Accuracy (RD and pCR) affected by
tumor pattern on MRI, histology type and grade, molecular subtypes and NAC regimen, targeting therapy
Understanding the factors affecting MRI accuracy in RD evaluation can help modify the clinical management for breast cancer patients after NAC.

35. Thank you!!