Eur Radiol (2005) 15: 1224–1233DOI 10.1007/s00330-005-2656-6 BREAST

Filippo MontemurroLaura MartincichGiovanni De RosaStefano CirilloVincenzo MarraNicoletta BigliaMarco GattiPiero SismondiMassimo AgliettaDaniele Regge

Received: 7 July 2004Accepted: 22 November 2004Published online: 27 January 2005# Springer-Verlag 2005

Dynamic contrast-enhanced MRIand sonography in patients receiving primarychemotherapy for breast cancer

Abstract We compared dynamiccontrast-enhanced MRI (DCE-MRI)and sonography (US) for monitoringtumour size in 21 patients with breastcancer undergoing primary chemo-therapy (PCT) followed by surgery.The correlation between DCE-MRIand US measurements of tumour size,defined as the product of the twomajordiameters, was 0.555 (P=0.009), 0.782(P<0.001), and 0.793 (P<0.001) atbaseline, and after two and four cyclesof PCT, respectively. The mediantumour size was significantly largerwhen measured by DCE-MRI than by

US at baseline (1472 vs 900 mm2,P<0.001) and after two cycles of PCT(600 vs 400 mm2, P=0.009). AfterPCT, themedian tumour sizemeasuredby the two techniques was similar (256vs 289 mm2 for DCE-MRI and US,respectively, P=0.859). The correla-tion with the histopathological majortumour diameter was 0.824 (P<0.001)and 0.705 (P<0.001) for post-treat-ment DCE-MRI and US, respectively.Measurements of the final majortumour diameter by DCE-MRI tendedto be more precise, including casesachieving a pathological complete re-sponse. Randomized trials are war-ranted to establish the clinical impactof the initial discrepancy in tumoursize estimates between DCE-MRI andUS, and the trend towards a betterdefinition of the final tumour sizeprovided by DCE-MRI in this clinicalsetting.

Keywords Sonography . Magneticresonance imaging . Breastneoplasms . Primary chemotherapy

This work was accepted as a scientific pre-sentation at the ECR 2004 meeting.

F. Montemurro . M. AgliettaUnit of Medical Oncology, Institute forCancer Research and Treatment (IRCC),Strada Provinciale 142,10060 Candiolo, Torino, Italy

L. Martincich . S. Cirillo . V. Marra .D. ReggeUnit of Diagnostic Imaging, Institute forCancer Research and Treatment (IRCC),Strada Provinciale 142,10060 Candiolo, Torino, Italy

G. De RosaUnit of Pathology, Institute for CancerResearch and Treatment (IRCC),Strada Provinciale 142,10060 Candiolo, Torino, Italy

N. Biglia . P. SismondiUnit of Gynecological Oncology,Institute for Cancer Research andTreatment (IRCC),Strada Provinciale 142,10060 Candiolo, Torino, Italy

M. GattiUnit of Radiotherapy, Institute forCancer Research and Treatment (IRCC),Strada Provinciale 142,10060 Candiolo, Torino, Italy

L. Martincich (*)Unit of Radiology, Institute for CancerResearch and Treatment (IRCC),Strada Provinciale 142,10060 Candiolo, Torino, Italye-mail: laura.martincich@ircc.itTel.: +39-11-9933328Fax: +39-11-9933301

Introduction

The use of primary chemotherapy (PCT) in women withlocally advanced or large breast cancer is well established[1]. Once limited to patients with inoperable tumours, thisapproach is now increasingly being used in patients withlarge operable tumours, with the aim of reducing the tu-mour size and, consequently, allowing less-extensive sur-gery [2]. The administration of chemotherapy before surgery

has the clinically relevant advantage of allowing an “invivo” assessment of chemosensitivity, by a closemonitoringof tumour shrinkage during the treatment. This has potentialimplications in treatment tailoring, by the identification ofpoor responders early in the course of the planned treat-ment, as well as in planning the extent of breast surgery.To date there is no agreement on the optimal tool for

tumour response monitoring in breast cancer patients un-dergoing PCT. Clinical palpation, sonography (US) and

mammography are widely utilized and cost effective [3].However, each shows limitations and inaccuracies, espe-cially in defining residual tumour after PCT [4–6]. Despitehigher costs and lower availability, magnetic resonanceimaging (MRI) is increasingly being used in several areas ofbreast cancer management [7–16]. In particular, dynamiccontrast-enhanced MRI (DCE-MRI) allows the estimationof both morphological and functional parameters related tomalignancy, resulting in potential improvements in theevaluation of patients during and upon completion of PCT[7, 12, 17–19]. Definition of tumour extent at diagnosis,detection of multicentricity/multifocality, evaluation of tu-mour shrinkage during PCT and of residual tumour beforesurgery and prediction of histopathological response areareas of the management of patients undergoing PCTwhereDCE-MRI has shown its potential.

Despite these features, at present DCE-MRI of the breastis still considered an adjunct modality, complementing, butnot replacing, US and mammography.

We designed a prospective evaluation of DCE-MRI andUS performed in parallel before, during and after PCT, inpatients with locally advanced breast cancer. Our main aimwas to study the correlation between tumour size evaluatedby DCE-MRI and US in this group of patients.

Materials and methods

Patients and treatment

A group of 21 women, median age 49 years (range 37–64years), with locally advanced or large operable breast can-cer, receiving PCT at our Institution, were monitored inparallel by DCE-MRI and US. Clinical stages by the TNMclassification [20] were: stage II with a major tumourdiameter ≥3 cm (ten patients), stage IIIA (four patients), andstage IIIB (seven patients). The seven patients staged as IIIBpresented with a single node and accompanying clinicalsigns of skin involvement (skin thickening, erythema, andpeau d’orange). Patients with inflammatory breast cancer(T4d) were excluded from this study. Chemotherapy for allthe patients was a combination of doxorubicin 50 mg/m2

bolus i.v. followed by paclitaxel 175 mg/m2 as a 3-h i.v.infusion, every 3 weeks for four cycles. Patients were re-quired to provide informed consent before entering thestudy.

Methods

Breast DCE-MRI and US were performed on the sameplanned day by two radiologists who were blinded to theresults of the other (L.M., V.M.). The examinations werecarried out at baseline, after two cycles of PCT, and uponcompletion of PCT within 1 week before surgery. A corebiopsy of the tumour was performed after the baseline

DCE-MRI of the breast in order to obtain adequate tumourtissue for histological diagnosis, determination of histo-logical grade and hormone receptor status. Similarly, forpatients with signs of skin involvement, confirmatorypunch biopsies were performed after the baseline DCE-MRI. Staging work-up was carried out according to in-stitutional guidelines.

Sonography

Sonography was performed using an AU5 ESAOTE equip-ment and a linear high-frequency transducer of 10/13 MHz,with the patient both in a supine and oblique position witharms above the head. For every lesion, the two majordiameters were obtained both in the transverse and sagittalplanes and measured by callipers.

DCE-MRI

Magnetic resonance images of the breast were acquiredwith a 1.5-T scanner using a dedicated surface multichannelcoil (GE Medical System, Milwaukee, Wis.) with the pa-tient in the prone position.

After an axial localizer, a morphological study was ob-tained using T1-weighted (TR 600 ms, TE 18.1 ms, slicethickness 5 mm, spacing 0.5 mm, FOV 24×24, matrix512×224, acquisition time 4.32 min) and FSEIR (TR 4400ms, TE 12.6 ms, TI 150 ms, slice thickness 5 mm, spacing0.5 mm, FOV 24×24, matrix 320×192, acquisition time3.36 min) sequences in the sagittal plane.

The dynamic study was performed using a 3D SPGR fat-suppressed sequence in the coronal plane, with the fol-lowing parameters: TR 8.9 ms, TE 4.2 ms, flip angle 20°,slice thickness 2.6 mm without interval, matrix 256×256and temporal resolution of about 60 s, according to thevolume of the breasts and to the FOV. The 3D sequencewas acquired before and five times continuously afterintravenous injection of gadolinium chelate (Magnevist,Schering); a late acquisition was obtained 2 min after thelast sequence. The acquisitions were started simultaneouslywith bolus injection of 0.1 mmol/kg of gadolinium chelate,infused into the antecubital vein by power injector, at arate of 2 ml/s and followed by a saline flush. The acquiredimages were transferred to a workstation (Advantage Win-dow 4.0, GE Medical Systems) for postprocessing, whichincluded image subtraction, MIP reconstruction and time/intensity (T/I) curves analysis.

A three-step process was used to ensure the accuracy ofDCE-MRI measurements of tumour lesions. First, subtrac-tion of basal images from postcontrast acquisitions wasperformed to improve visualization of areas of contrastuptake. For the evaluation of tumour margins we chosesubtracted images from the third postcontrast acquisition(about 180 s following the start of contrast administration)

1225

as suggested by Cheung et al. [21]. On the basis of thesubtracted images, MIP reconstructions were obtained inthe three orthogonal planes. A semiquantitative analysis ofregions of interest (ROI) was then carried on in order tocorrectly identify areas of active tumour tissue at baseline,and during and after chemotherapy. A synthesis of dynamicand morphological features (including pattern of contrastuptake and kinetic characteristics) was evaluated. In par-ticular, T/I curves were drawn for each of the enhancingareas and a >50% increase of early signal intensity wasinterpreted as a sign for malignancy [14]. Finally, the twomajor diameters of the suspicious areas were measuredusing electronic callipers onMIP reconstructions. TheDCE-MRI image interpretation and measurements were per-formed independentlyby twoexperienced radiologists (L.M.and S.C.).

Tumour response and histopathological evaluation

Tumour response at the end of PCT was evaluated bycomparing tumour size estimated by DCE-MRI and US atbaseline and upon completion of PCT. Complete response(CR) was defined as the disappearance of any evidence ofdisease, partial response (PR) was defined as a >50% re-duction in tumour size, stable disease (SD) was defined aseither a ≤50% reduction or a ≤25% increase in tumour size,and progressive disease (PD) was defined as a >25% in-crease in tumour size.

All the patients were evaluated at baseline and uponcompletion of PCT by a multidisciplinary breast team. Thebaseline and posttreatment imaging studies were reportedby radiologists and presented to the multidisciplinary teamfor decision on patient management. With the exception ofthe seven stage IIIB patients, breast conserving surgery wasoffered to patients whose posttreatment target tumour vol-ume, estimated on the basis of both US and DCE-MRI, wasdeemed to be resectable with free margins [1].

After surgery, breast specimens were handled accordingto the hospital routines. Histopathological examination in-cluded macroscopic localization of the lesions in the spec-imens, fixation in formalin and embedding in paraffinbefore large-size sections were made. Cross-sections of thespecimens were made at 5-mm intervals perpendicular tothe line connecting the nipple and the centre of the tumour.Additional sections of grossly normal breast tissue wereprepared in patients who had undergone mastectomy orquadrantectomy.

Specimens were examined microscopically usinghaematoxylin and eosin staining. In order to facilitate com-parisons, we chose to measure only the largest histopath-ological diameter (including also the intraductal tumourcomponent if present) as representative of the final tumoursize, which was compared with the largest tumour diametermeasured by the posttreatment DCE-MRI and US. Whenmultiple tumour foci were present at histopathology, the

largest diameter of the area including the tumour foci wasconsidered.

The histopathological response to chemotherapy wasevaluated using a five-point assessment scheme describedby Smith et al. [22]: grade 1, some alteration to individualmalignant cells but no reduction in overall numbers ascompared with the pretreatment core biopsy; grade 2, a mildloss of invasive tumour cells but overall cellularity stillhigh; grade 3, a considerable reduction in tumour cells upto an estimated 90% loss; grade 4, a marked disappearanceof invasive tumour cells such that only small clusters ofwidely dispersed cells could be detected; and grade 5, noinvasive tumour cells identifiable in the sections from thesite of the previous tumour, that is, only in situ disease ortumour stroma remaining. A grade 5 response was deemedto represent a pathological complete response (pCR) of theprimary cancer.

Statistical analyses

For each of the identified lesions, the size was defined asthe product of the two major diameters. Wilcoxon’s signed-ranks test was used to evaluate the statistical significancebetween the differences in size measured by DCE-MRI andUS at each time-point. Spearman’s correlation coefficientwas determined to evaluate the correlation between the sizeof the lesions as measured by DCE-MRI and US, and thecorrelation between the histopathological major tumourdiameter and the major tumour diameter as measured byboth DCE-MRI and US on completion of PCT.

Results

Baseline examinations

Both baseline DCE-MRI and US detected 21 breast lesions(example shown in Fig. 1). No patient showed evidenceof multicentric or contralateral breast cancer. Core biop-sies showed infiltrating ductal carcinoma in 20 patients andinfiltrating lobular carcinoma in 1 patient. At MRI, all thelesions appeared as a focal mass with irregular margins,characterized by isohypointensity or isohyperintensity onthe T1-weighed and FSEIR images, respectively. After con-trast medium administration, 11 lesions showed homoge-neous enhancement and the remaining 10 showed peripheralenhancement with the ring sign. The median tumour sizewas 1472mm2 (range 360–8760mm2) and 900 mm2 (range225–2520 mm2) for DCE-MRI and US, respectively (P<0.001). The correlation between DCE-MRI and US mea-surements of tumour size was 0.555 (P=0.009, Fig. 2a).DCE-MRI andUSmeasurements were identical in 5 lesions(Table 1). The other 16 lesions, including all 7 stage IIIBtumours, were larger when measured by DCE-MRI than byUS (difference range 135–7860 mm2).

1226

Intermediate examinations

After two cycles of chemotherapy, the lesions detected wereagain 21 for both DCE-MRI and US (example shown inFig. 3). No lesions showed changes of the morphodynamicDCE-MRI pattern.

The median tumour size was 600 mm2 (range 49–3551mm2) and 400 mm2 (range 42–1600 mm2) for DCE-MRIand US, respectively (P=0.009). The correlation betweenDCE-MRI and US measurements of tumour size was 0.782(P<0.001, Fig. 2b).

DCE-MRI and US measurements were identical in twolesions (Table 1). Five lesions, including one presentingwith accompanying signs of inflammation that had re-gressed after two cycles of PCT, were smaller whenmeasured by DCE-MRI than by US (difference range 51–215 mm2). The remaining 14 lesions identified by both

techniques appeared larger when measured by DCE-MRIthan by US (difference range 60–2875 mm2). Four of thesix remaining stage IIIB tumours showed the largest dif-ference in measurement between the two methods (differ-ence range 2,090–2,785 mm2).

After treatment

Upon completion of four cycles of chemotherapy the num-bers of lesions identified by DCE-MRI and US were 18 and21, respectively (example shown in Fig. 4).

The median tumour size was 256 mm2 (range 0–1600mm2) and 289 mm2 (range 16–600mm2) for DCE-MRI andUS, respectively (P=0.859). The correlation between DCE-MRI and US measurements of tumour size was 0.793(P<0.001, Fig. 2c). DCE-MRI and US measurements were

Fig. 1 Baseline examination: at MRI, MIP reconstructions (a, c)show a round-shaped area of homogeneous enhancement in theexternal quadrants of the left breast; at US (b, d) the lesion appears asa hypoechogenic area with irregular margins. Tumour size was

evaluated by multiplying the two major diameters of the lesion. In thecase displayed in the figure the MRI estimate of tumour size waslarger than the US estimate (2177 and 823 mm2 for MRI and US,respectively)

1227

identical in 11 lesions (Table 1). Three lesions were nolonger identifiable by DCE-MRI, but still visible by US.Four other lesions, including two stage IIIB tumours, werelarger when measured by DCE-MRI than by US (difference

range 162–871 mm2). Three lesions were smaller whenmeasured by DCE-MRI than US (difference range 75–500mm2).

Tumour response and histopathological evaluation

Upon completion of PCT, tumour responses according toUS were as follows: 15 patients achieved a PR (71%), and 6achieved SD (29%). Tumour responses according to DCE-MRI were as follows: three patients achieved a CR (14%),15 achieved a PR (71%), and three achieved SD (14%). Thethree patients identified as showing a CR by DCE-MRIwere classified as PR by US.

After PCT all the patients underwent breast surgery,including five who were not considered initially operable.Surgery consisted of quadrantectomy or lumpectomy in 5

a

-100400900

1400190024002900340039004400490054005900640069007400790084008900

-100 400 900 1400 1900 2400 2900

MR

I pro

duct

of m

ajor

dia

met

ers

(mm

2 )

b

-2000

200400600800

100012001400160018002000220024002600280030003200340036003800

-200 0 200 400 600 800 1000 1200 1400 1600 1800 2000

US product of major diameters (mm2)

MR

I pro

duct

of m

ajor

dia

met

ers

(mm

2 )

c

-100

100

300

500

700

900

1100

1300

1500

1700

1900

-100 100 300 500 700 900 1100 1300 1500 1700 1900

MR

I pro

duct

of m

ajor

dia

met

ers

(mm

2 )

US product of major diameters (mm2)

US product of major diameters (mm2)

3Fig. 2 a Correlation between baseline US (x-axis) and DCE-MRI (y-axis) measurements of tumour size. Spearman’s correlation coeffi-cient was 0.555 (P=0.009). b Correlation between US (x-axis) andDCE-MRI (y-axis) measurements of tumour size after two cycles ofchemotherapy. Spearman’s correlation coefficient was 0.782 (P<0.001). c Correlation between US (x-axis) and DCE-MRI (y-axis)measurements of tumour size after treatment. Spearman’s correlationcoefficient was 0.793 (P<0.001)

Table 1 Percentage difference in tumour size between DCE-MRIand US (positive values indicate larger tumours at DCE-MRI,negative values indicate larger tumours at US)

Patient ID Baseline (%) Intermediate (%) Posttreatment (%)

1a +37 −56 −752 0 +42 03 0 +29 04 +28 −51 −100b

5 +16 −26 06 0 +61 07 +25 −40 −100b

8a +36 +24 09 +37 0 −100b

10 +64 +72 011 +59 +13 012a +11 +19 +8213 0 0 014 +49 +7 015a +60 +73 +5416a +79 +80 017a +65 +65 −3118a +90 +87 019 +54 +69 +8220 0 −19 +4621 +39 +44 −93aStage IIIB tumours, presenting as single node with accompanyingsigns of skin involvementbPathological complete responders

1228

patients and modified radical mastectomy in the remaining16 patients. Three patients who had become eligible forbreast-conserving surgery as a result of PCT opted formodified radical mastectomy.

Histopathological responses according to the adoptedscoring system were grade 5 in three patients, grade 4 infour patients, grade 3 in four patients, grade 2 in four pa-tients and grade 1 in six patients. Accompanying ductalcarcinoma in situ was present in three specimens. Majortumour diameter by both DCE-MRI and US at the end oftreatment showed significant correlation with the histo-pathological major tumour diameter of the lesions. The cor-

relation coefficient was 0.824 (P<0.001) and 0.705 (P<0.001) for DCE-MRI and US, respectively (Fig. 5).

Figure 6 shows the difference in major tumour diameterafter PCT between DCE-MRI and US compared with thehistopathological major tumour diameter. Despite a lack ofstatistically significant difference between the two methods(Wilcoxon’s signed-ranks test, P=0.506), DCE-MRI mea-surements tended to be more precise (Fig. 6). The threepatients with pCR were correctly identified by DCE-MRI,but not by US. In two patients DCE-MRI provided a largeoverestimate of the residual tumour diameter (Fig. 6). Themajor tumour diameters in the two patients by DCE-MRIwere 15 and 25 mm, respectively, whereas the major his-

Fig. 3 Same case as Fig. 1. Intermediate examination: after two cycles of PCT the size of the lesion is reduced and the MRI and US estimatesstill differ, although the degree of discrepancy is much lower than at baseline (933 vs 780 mm2 for MRI and US, respectively)

1229

topathological diameters were 2 and 1 mm, respectively.Both patients showed a histopathological grade 4 responseto PCT and had accompanying high-grade in situ ductalcarcinoma.

In the only patient with infiltrating lobular carcinoma inour series, DCE-MRI provided a large underestimate oftumour diameter (10 vs 22 mm for DCE-MRI and his-topathology, respectively). For both DCE-MRI and US, thecorrelations with histopathological diameter were differentaccording to the degree of tumour regression. In the 7 pa-tients with pronounced tumour regression (grade 4 and 5histopathological response) the correlation coefficientswere 0.731 (P=0.06) and 0.113 (P=0.81) for DCE-MRIand US, respectively, whereas in the 14 patients with lesser

or no tumour regression (grades 1, 2 and 3 histopathologicalresponse), the correlation coefficients were 0.951 (P<0.01)and 0.835 (P<0.01) for DCE-MRI and US, respectively.

Discussion

In the current study we were interested in comparing DCE-MRI and US as tools for monitoring tumour size in womenwith locally advanced breast cancer undergoing PCT. US,together with clinical palpation, is considered by many thegold standard for monitoring tumour size during PCT [3].Moreover, US is cost-effective and usually well accepted bythe patient. MRI techniques are less widely available, more

Fig. 4 Same case as Figs. 1 and 2. Presurgical examination: on completion of PCT both MRI (a, c) and US (b, d) show a further tumour sizereduction. MRI estimate of tumour size was calculated measuring the enhancing area suspicious for residual disease on the basis ofsemiquantitative analysis

1230

costly and carry a certain degree of discomfort for thepatient. For these reasons, at present MRI is considered anadjunct modality in patients undergoing PCT. In this clin-ical setting, an ideal diagnostic tool should provide anaccurate estimation of the baseline tumour extent and size,a correct evaluation of tumour shrinkage as a result of thetreatment and a precise evaluation of residual tumour, in-cluding identification of pathological complete responders.All these features are critical in planning the correct surgeryin order to minimize the local relapse rate [23].

At baseline DCE-MRI measurements of tumour sizewere significantly larger than US measurements; in fact,only 5 lesions had an identical tumour size by both tech-

niques, whereas the remaining 16were largerwhenmeasuredby DCE-MRI than US. As a result, although significant, thecorrelation between the twomethods was low (Fig. 2a). Thelargest discrepancy in lesion size was seen in the stage IIIBpatients, in whom the tumour node is accompanied bysigns of skin involvement (Table 1). The high sensitivity ofDCE-MRI in identifying malignant diffuse tissue infiltra-tion may account for this observation [24].

The interesting finding of our study is that the correlationbetween tumour size measured by DCE-MRI and by USimproved after two cycles and after completion of PCT(Fig. 2b, c). At the latter time-point, the median tumour sizemeasured by the two methods was similar (256 and 289mm2 for DCE-MRI and US, respectively, P=0.859). Thus,according to our results, DCE-MRI and US provide similarestimates of the extent of residual tumour before surgery.In general, in patients receiving PCT, the target volumeof resection is deemed to be the postchemotherapy ab-normality and attempts to remove the initial volume ofdisease are not considered necessary [1]. From this per-spective, according to our findings, the systematic use ofDCE-MRI would offer no additional advantage over USin planning surgery upon completion of PCT.

The point to be addressed is whether a potential under-estimation by US or, by contrast, a potential overestimationby DCE-MRI of the initial tumour size would affect apatient’s eligibility for PCT or the surgical choices upontreatment completion. Studies comparing MRI and US andmammography in breast cancer patients undergoing surgerywithout PCT suggest that, with respect to MRI, the other

a

-5

0

5

10

15

20

25

30

35

40

45

-5 0 5 10 15 20 25 30 35 40 45

Major histopathological diameter (mm)

MR

I maj

or d

iam

eter

(m

m)

b

-5

0

5

10

15

20

25

30

35

40

-5 0 5 10 15 20 25 30 35 40 45

Major histopathological diameter (mm)

US

maj

or d

iam

eter

(m

m)

Fig. 5 a Correlation between the histopathological major diameter(x-axis) and the major tumour diameter measured by DCE-MRI (y-axis) after completion of treatment. Spearman’s correlation coeffi-cient was 0.824 (P<0.001). The three patients with pCR were nolonger identifiable at DCE-MRI. b Correlation between the histo-pathological major diameter (x-axis) and the major tumour diametermeasured by US (y-axis) after completion of treatment. Spearman’scorrelation coefficient was 0.705 (P<0.001). Sonography detectedabnormalities indicating persistence of disease in all three patientswith pCR

Fig. 6 Difference in major tumour diameter at DCE-MRI and USperformed upon completion of PCT, compared with histopathologicaldiameter. The boxes represent the measured differences between the25th and the 75th percentiles. Outliers (open circles) are defined asdifferences that are outside the box at distances 1.5–3 times the size ofthe box, and extremes (asterisks) are defined as differences that areoutside the box at distances more than 3 times the size of the box

1231

techniques tend to underestimate the extent of disease [8,25, 26]. By contrast, in a recent study, Hata et al. found thatcompared with US or mammography, DCE-MRI tended tooverestimate the histopathological extent of disease in38.9% of the cases [27]. If surgical choices had been madeon the basis of DCE-MRI in that study, a number of un-necessary mastectomies would have been performed.Whilein the preoperative setting the use of MRI is still contro-versial, considering also the excellent rates of recurrencewhen breast-conserving surgery is guided by conventionalmethods of tumour size assessment [28], in the setting ofPCT additional consideration should be made. The feasi-bility of breast-conserving surgery depends on the extent oftumour regression as a result of the medical treatment.

In the National Surgical Adjuvant Breast and BowelProject (NSABP) study B18 four cycles of doxorubicinand cyclophosphamide (AC, 60 and 600 mg/m2 every 3weeks) as adjuvant (postsurgical) or neoadjuvant (PCT)treatment in women with operable breast cancer were com-pared [29]. The updated results of this study show higherrates of local relapse in patients initially deemed to requiremastectomy and who became eligible for breast-conservingsurgery after PCT, compared with patients who were ini-tially eligible for breast-conserving surgery [30]. This find-ing indicates that the initial evaluation of tumour extent is acritical step in the eligibility and management of womenundergoing PCT. For this reason we believe that our find-ings merit further evaluation in order to study their impacton surgical choices and patient outcomes.

Similar to the findings of other groups, we found thatDCE-MRI offers at least two potential advantages over US.First, despite a lack of statistical significance of the dif-ference in major residual tumour diameter by DCE-MRIand US compared with the histopathological major diam-eter, DCE-MRI measurements appeared to be more precise(Fig. 6) and the correlation coefficient was higher. Second,DCE-MRI correctly identified the three patients with noresidual tumour (pCR), whereas US failed to do so. Withregard to the first issue, as previously reported by otherauthors [19, 31], regressive phenomena induced by che-motherapy may limit the accuracy of DCE-MRI in thissetting. In fact, we also found a reduced correlation between

DCE-MRI and histopathological measures of the majortumour diameter according to the extent of tumour regres-sion (0.731 vs and 0.951 in patients with and withoutextensive tumour regression). However, although the smallnumbers involved limit the worth of this observation, USseemed to suffer even more from this limitation (correlationcoefficient in patients with extensive tumour regression0.113). Moreover, in contrast to other reports in the medicalliterature, we found no “false-positive results” (patientsclassified as complete responders with persistent disease athistopathology) among patients whose DCE-MRI docu-mented a complete response.

The achievement of a pCR has a recognized role inpredicting a favourable clinical outcome [32]. Current PCTstrategies are aimed at increasing the pathological responserate, a target that can be achieved using sequences of dif-ferent active chemotherapy regimens [22, 33]. Moreover,the identification of early responders or nonresponders iscrucial for correct therapeutic planning. We did not addressthe predictive value of tumour measurements by DCE-MRIor US during PCTwith respect to the achievement of a pCRin this study. However, our previous work and similar ex-periences from other groups suggest that morphological andfunctional information obtained using DCE-MRI early dur-ing the course of PCT results in high predictive value withrespect to the achievement of a pCR [17, 21, 34].

In summary, our study showed that before PCT, DCE-MRI and US provide significantly different estimates oftumour extent, with larger tumour size by DCE-MRI in 16out of 21 patients. At the completion of PCT, both tech-niques provided similar estimates of the major histopath-ological tumour diameter, with a trend towards a betterperformance of DCE-MRI. Although our findings suggestthat DCE-MRI may have a role in the management ofbreast cancer patients undergoing PCT, at present its rou-tine use in this clinical setting cannot be recommended.Adequately designed randomized clinical trials are neededto evaluate whether the potential advantages of DCE-MRIover conventional imaging really translate into a clinicalbenefit that overweighs the increased demands and costsof this technique.

References

1. Buchholz TA, Hunt KK, Whitman GJ,Sahin AA, Hortobagyi GN (2003)Neoadjuvant chemotherapy for breastcarcinoma: multidisciplinary considera-tions of benefits and risks. Cancer98:1150–1160. DOI 10.1002/cncr.11603

2. Kaufmann M, von Minckwitz G, SmithR, et al (2003) International expert panelon the use of primary (preoperative)systemic treatment of operable breastcancer: review and recommendations. JClin Oncol 21:2600–2608. DOI10.1200/JCO.2003.01.136

3. Herrada J, Iyer RB, Atkinson EN,Sneige N, Buzdar AU, Hortobagyi GN(1997) Relative value of physical ex-amination, mammography, and breastsonography in evaluating the size of theprimary tumor and regional lymph nodemetastases in women receiving neoad-juvant chemotherapy for locally ad-vanced breast carcinoma. Clin CancerRes 3:1565–1569

1232

4. Vinnicombe SJ,MacVicar AD, Guy RL,Sloane JP, Powles TJ, Knee G, HusbandJE (1996) Primary breast cancer: mam-mographic changes after neoadjuvantchemotherapy, with pathologic correla-tion. Radiology 198:333–340

5. Huber S, Wagner M, Zuna I, Medl M,Czembirek H, Delorme S (2000)Locally advanced breast carcinoma:evaluation of mammography in theprediction of residual disease after in-duction chemotherapy. Anticancer Res20:553–558

6. Cocconi G, Di Blasio B, Alberti G,Bisagni G, Botti E, Peracchia G (1984)Problems in evaluating response ofprimary breast cancer to systemic ther-apy. Breast Cancer Res Treat 4:309–313

7. Sardanelli F, Lozzelli A, Fausto A(2003) MR imaging of the breast:indications, established technique,and new directions. Eur Radiol 13[Suppl 3]:N28–N36

8. Esserman L, Hylton N, Yassa L, BarclayJ, Frankel S, Sickles E (1999) Utility ofmagnetic resonance imaging in themanagement of breast cancer: evidencefor improved preoperative staging. JClin Oncol 17:110–119

9. Szabo BK, Aspelin P, KristoffersenWM, Tot T, Bone B (2003) Invasivebreast cancer: correlation of dynamicMR features with prognostic factors.Eur Radiol 13:2425–2435. DOI10.1007/s00330-003-2000-y

10. Fischer DR, Baltzer P, Malich A,Wurdinger S, Freesmeyer MG, Marx C,Kaiser WA (2004) Is the “bloomingsign” a promising additional tool todetermine malignancy in MR mam-mography? Eur Radiol 14:394–401.DOI 10.1007/s00330-003-2055-9

11. KristoffersenWM,Aspelin P, SylvanM,Bone B (2003) Comparison of lesionsize estimated by dynamic MR imaging,mammography and histopathology inbreast neoplasms. Eur Radiol 13:1207–1212. DOI 10.1007/s00330-002-1718-2

12. Walter C, Scheidhauer K, Scharl A, et al(2003) Clinical and diagnostic value ofpreoperative MR mammography andFDG-PET in suspicious breast lesions.Eur Radiol 13:1651–1656. DOI10.1007/s00330-002-1816-1

13. Viehweg P, Rotter K, Laniado M,Lampe D, Buchmann J, Kolbl H,Heywang-Kobrunner S (2004) MR im-aging of the contralateral breast inpatients after breast-conserving therapy.Eur Radiol 14:402–408. DOI 10.1007/s00330-003-2086-2

14. Goerres GW, Michel SC, Fehr MK,et al (2003) Follow-up of women withbreast cancer: comparison betweenMRIand FDG PET. Eur Radiol 13:1635–1644. DOI 10.1007/s00330-002-1720-8

15. Rieber A, Schramm K, Helms G,et al (2003) Breast-conserving surgeryand autogenous tissue reconstructionin patients with breast cancer: efficacyof MRI of the breast in the detectionof recurrent disease. Eur Radiol 13:780–787. DOI 10.1007/s00330-002-1538-4

16. Abraham DC, Jones RC, Jones SE, et al(1996) Evaluation of neoadjuvantchemotherapeutic response of locallyadvanced breast cancer by magneticresonance imaging. Cancer 78:91–100

17. Martincich L, Montemurro F, De RosaG, et al (2004) Monitoring response toprimary chemotherapy in breast cancerusing dynamic contrast-enhanced mag-netic resonance imaging. Breast CancerRes Treat 83:67–76

18. Wasser K, Klein SK, Fink C, et al (2003)Evaluation of neoadjuvant chemothera-peutic response of breast cancer usingdynamic MRI with high temporal reso-lution. Eur Radiol 13:80–87

19. Rieber A, Brambs HJ, Gabelmann A,Heilmann V, Kreienberg R, Kuhn T(2002) Breast MRI for monitoring re-sponse of primary breast cancer to neo-adjuvant chemotherapy. Eur Radiol12:1711–1719. DOI 10.1007/s00330-001-1233-x

20. Singletary SE, Allred C, Ashley P, et al(2002) Revision of the American JointCommittee on Cancer staging systemfor breast cancer. J Clin Oncol 20:3628–3636

21. Cheung YC, Chen SC, Su MY, et al(2003) Monitoring the size and responseof locally advanced breast cancers toneoadjuvant chemotherapy (weekly pa-clitaxel and epirubicin) with serial en-hanced MRI. Breast Cancer Res Treat78:51–58

22. Smith IC, Heys SD, Hutcheon AW, et al(2002) Neoadjuvant chemotherapy inbreast cancer: significantly enhancedresponse with docetaxel. J Clin Oncol20:1456–1466

23. Kuerer HM, Singletary SE, Buzdar AU,et al (2001) Surgical conservation plan-ning after neoadjuvant chemotherapyfor stage II and operable stage III breastcarcinoma. Am J Surg 182:601–608

24. Esserman L, Kaplan E, Partridge S, et al(2001) MRI phenotype is associatedwith response to doxorubicin and cy-clophosphamide neoadjuvant chemo-therapy in stage III breast cancer. AnnSurg Oncol 8:549–559

25. Davis PL, Staiger MJ, Harris KB,Ganott MA, Klementaviciene J,McCarty KS Jr, Tobon H (1996) Breastcancer measurements with magneticresonance imaging, ultrasonography,and mammography. Breast Cancer ResTreat 37:1–9

26. Boetes C, Mus RD, Holland R, et al(1995) Breast tumors: comparative ac-curacy of MR imaging relative tomammography and US for demonstrat-ing extent. Radiology 197:743–747

27. Hata T, Takahashi H, Watanabe K,Takahashi M, Taguchi K, Itoh T, TodoS (2004) Magnetic resonance imagingfor preoperative evaluation of breastcancer: a comparative study withmammography and ultrasonography.J Am Coll Surg 198:190–197. DOI10.1016/j.jamcollsurg.2003.10.008

28. Morrow M (2004) Magnetic resonanceimaging in the preoperative evaluationof breast cancer: primum non nocere. JAm Coll Surg 198:240–241. DOI10.1016/j.jamcollsurg.2003.10.013

29. Fisher B, Bryant J, Wolmark N, et al(1998) Effect of preoperative chemo-therapy on the outcome of women withoperable breast cancer. J Clin Oncol16:2672–2685

30. Wolmark N, Wang J, Mamounas E,Bryant J, Fisher B (2001) Preoperativechemotherapy in patients with operablebreast cancer: nine-year results from theNational Surgical Adjuvant Breast andBowel Project B-18. J Natl Cancer InstMonogr 30:96–102

31. Wasser K, Sinn HP, Fink C, et al (2003)Accuracy of tumor size measurement inbreast cancer using MRI is influencedby histological regression induced byneoadjuvant chemotherapy. Eur Radiol13:1213–1223

32. Kuerer HM, Newman LA, Smith TL,et al (1999) Clinical course of breastcancer patients with complete patho-logic primary tumor and axillary lymphnode response to doxorubicin-basedneoadjuvant chemotherapy. J ClinOncol 17:460–469

33. Bear HD, Anderson S, Brown A, et al(2003) The effect on tumor response ofadding sequential preoperative doce-taxel to preoperative doxorubicin andcyclophosphamide: preliminary resultsfrom National Surgical Adjuvant Breastand Bowel Project Protocol B-27. J ClinOncol 21:4165–4174. DOI 10.1200/JCO.2003.12.005

34. Warren RM, Bobrow LG, Earl HM,et al (2004) Can breast MRI help in themanagement of women with breastcancer treated by neoadjuvant chemo-therapy? Br J Cancer 90:1349–1360

1233