Note: This copy is for your personal non-commercial use only. To order presentation-ready copies for distribution to your colleagues or clients, contact us at www.rsna.org/rsnarights.

Original Research  n  Breast

Ann Yi, MD Nariya Cho, MD Kyung-Sook Yang, PhD Wonshik Han, MD Dong-Young Noh, MD Woo Kyung Moon, MD

Imaging

Breast Cancer Recurrence in Patients with Newly Diagnosed Breast Cancer without and with Preoperative MR Imaging: A Matched Cohort Study1 Purpose:

To compare breast cancer disease-free survival (DFS) outcomes of patients with newly diagnosed breast cancer without and with preoperative magnetic resonance (MR) imaging.

Materials and Methods:

This study was approved by an institutional review board, and informed consent was waived. From 2004 to 2009 (unilateral MR imaging from 2004 to 2006 vs bilateral MR imaging from 2007 to 2009), patients with breast cancer without preoperative MR imaging (no MR imaging group) were matched with those with preoperative MR imaging (MR imaging group) according to age, histologic grade, nuclear grade, tumor size, nodal status, stage, hormone receptor status, Ki-67 status, molecular subtype, and lymphovascular invasion. Survival analysis was performed by using Kaplan-Meier estimates. A marginal model was used to evaluate the effect of preoperative MR imaging on DFS.

Results:

A total of 371 patient pairs from the unilateral imaging period and 97 patient pairs from the bilateral imaging period were matched. During the unilateral imaging period, the MR imaging group had better local-regional recurrence DFS (hazard ratio [HR], 0.33; 95% confidence interval [CI]: 0.12, 0.91; P = .032) than did the no MR imaging group; however, no difference was found for contralateral breast (P = .440) or distant recurrence (P = .515) DFS. During the bilateral imaging period, the MR imaging group had better contralateral breast cancer DFS (HR, 0.03; 95% CI: 0.04, 0.21; P , .001) than the no MR imaging group; however, no difference was found for localregional (P = .180) or distant recurrence (P = .178) DFS.

Conclusion:

Preoperative bilateral breast MR imaging for staging of breast cancer was associated with a reduced risk of contralateral breast recurrence; however, no observed reduction in risk of local-regional or distant recurrence was shown.

1

 From the Department of Radiology, Seoul National University Hospital, 101 Daehak-ro, Jongno-gu, Seoul 110-744, Republic of Korea (A.Y., N.C., W.K.M.); Department of Radiology, Seoul National College of Medicine, Seoul, Republic of Korea (N.C., W.K.M.); Institute of Radiation Medicine, Seoul National University Medical Research Center, Seoul, Republic of Korea (N.C., W.K.M.); Department of Biostatistics, Korea University College of Medicine, Seoul, Republic of Korea (K.S.Y.); and Department of Surgery, Seoul National University College of Medicine, Seoul National University Hospital, Seoul, Republic of Korea (W.H., D.Y.N.). Received September 15, 2014; revision requested October 30; revision received January 30, 2015; accepted February 6; final version accepted February 12. Address correspondence to N.C. (e-mail: [email protected]).

 RSNA, 2015

q

Online supplemental material is available for this article.

 RSNA, 2015

q

Radiology: Volume 276: Number 3—September 2015  n  radiology.rsna.org

695

BREAST IMAGING: Breast Cancer Recurrence in Matched Cohort Groups

I

t has been suggested that preoperative magnetic resonance (MR) imaging should not be routinely used for staging of disease in women with newly diagnosed breast cancers. The main reason for this suggestion was that it does not reduce the risk of local or distant recurrence or the reoperation rate or rate of conversion from conservation therapy to mastectomy, although studies have shown that 16%–20% of additional diseases can be depicted by using MR imaging (1– 5). This suggestion is also well in line with studies showing that long-term survival outcomes of breast conservation therapy are equivalent to those of mastectomy (6,7). In addition, the fact that multiple cancers not visible at mammography can be well treated by using radiation and systemic therapy

Advances in Knowledge nn Preoperative bilateral breast MR imaging was associated with 85% reduction in risk of recurrence for total recurrences (hazard ratio [HR], 0.15; 95% confidence interval [CI]: 0.07, 0.32; P , .001), mainly from contralateral breast recurrence risk reduction (HR, 0.03; 95% CI: 0.04, 0.21; P , .001). nn Patients with breast cancer who had undergone preoperative MR imaging had better total recurrence and contralateral breast recurrence disease-free survival than those who had not undergone preoperative MR imaging in the bilateral imaging period, whereas no differences were found between the two groups in the unilateral imaging period. nn Of the 22 metachronous contralateral breast cancers from the bilateral imaging period, only one node-negative pT1 invasive cancer (one [4.5%] of 22) was detected in the MR imaging group, whereas the remaining 21 cancers detected in the no MR imaging group varied from stage 0 to III, 38.0% (eight of 21) of which were node positive. 696

Yi et al

has further limited the role of preoperative MR imaging for the ipsilateral breast (8). However, the role of preoperative MR imaging for the contralateral breast of women with newly diagnosed breast cancer remains controversial. Two recent retrospective studies (9,10) reported that metachronous contralateral breast cancer incidence can be decreased in women who had undergone preoperative MR imaging. To the contrary, other studies (11,12) have reported that contralateral breast cancer rates were not different between MR imaging and non–MR imaging groups. As these previous studies have been criticized for nonequivalent comparison groups and use of older MR imaging techniques, better structured studies using the newest techniques is warranted. Thus, we hypothesized that a matched-cohort analysis may be able to control nonequivalent clinical-pathologic variables affecting disease-free survival (DFS) outcomes and that separate analysis according to the MR imaging technique used could clearly demonstrate the effect of MR imaging on contralateral breast recurrence outcomes. Therefore, the purpose of our study was to investigate the effect of preoperative breast MR imaging on DFS outcomes with special emphasis on contralateral breast recurrences in patients with newly diagnosed breast cancer by using concurrently matched cohorts according to the use of unilateral versus bilateral breast MR techniques.

Materials and Methods Approval for this retrospective study was obtained from the institutional

Implication for Patient Care nn Preoperative bilateral breast MR imaging is associated with reduced metachronous contralateral breast cancer incidence, although no observed reduction in risk of local-regional or distant recurrence was shown in patients with newly diagnosed breast cancer.

review board. Requirement of informed consent was waived.

Study Cohort Between January 2004 and March 2009, 4400 consecutive women with newly diagnosed breast cancers who underwent curative surgery were identified from the Breast Imaging Center database of Seoul National University Hospital. Among them, we excluded women who had undergone neoadjuvant chemotherapy (n = 145), had a past history of breast cancer (n = 72), had a past history of synchronous bilateral breast cancer (n = 65), had a past history of metastatic disease at presentation (n = 64), had no availability of 6-month follow-up data (n = 46), and had no availability of immunohistochemical information (n = 24). Among the remaining 3984 patients (median age, 48.5 years; age range, 20–89 years) with newly diagnosed breast cancers, 3440 patients had undergone preoperative MR imaging and 544 patients had not (Table E1 [online]). A total of 3094 of the 3984 patients had been previously reported (10). The prior study dealt with the role of preoperative MR imaging in the contralateral breast by using a historical controlled design, and it did not include a group that had not undergone MR imaging. For each woman with breast

Published online before print 10.1148/radiol.2015142101  Content codes: Radiology 2015; 276:695–705 Abbreviations: CI = confidence interval DFS = disease-free survival HER2 = human epidermal growth factor receptor type 2 HR = hazard ratio Author contributions: Guarantor of integrity of entire study, N.C.; study concepts/ study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; agrees to ensure any questions related to the work are appropriately resolved, all authors; literature research, A.Y., N.C., W.K.M.; clinical studies, A.Y., N.C., K.S.Y., W.H., D.Y.N., W.K.M.; statistical analysis, A.Y., N.C., K.S.Y.; and manuscript editing, A.Y., N.C. Conflicts of interest are listed at the end of this article.

radiology.rsna.org  n Radiology: Volume 276: Number 3—September 2015

BREAST IMAGING: Breast Cancer Recurrence in Matched Cohort Groups

cancer who had not undergone preoperative MR imaging (no MR imaging group), 11 covariates were matched to women who had undergone preoperative MR imaging (MR imaging group) in terms of age (,45 years or 45 years), histologic grade (I or II; III), nuclear grade (I or II; III), pathologic tumor size (20 mm or .20 mm), pathologic node status (negative or positive), pathologic stage (0 or I; II or III), estrogen receptor status (negative or positive), progesterone receptor status (negative or positive), Ki-67 status (.14% or 14%), molecular subtype (luminal, HER2 [human epidermal growth factor receptor type 2]–enriched, or triple negative), and lymphovascular invasion (no or yes). Since 2007, owing to advanced MR imaging techniques with faster acquisitions, the unilateral breast MR imaging protocol of our institution was able to be converted to a bilateral breast MR imaging protocol (13). Thus, unilateral MR imaging was performed between 2004 and 2006 and bilateral MR imaging was performed between 2007 and 2009.

Preoperative Evaluation For preoperative staging of breast cancers, clinical breast examinations, bilateral mammography, bilateral breast ultrasonography (US), and unilateral (during unilateral period) or bilateral (during bilateral period) MR imaging were performed. Patients who had not undergone preoperative MR imaging, in both unilateral and bilateral imaging periods, underwent bilateral mammography and bilateral US. All MR examinations were performed by using a 1.5-T system and a dedicated breast coil. Between 2004 and 2006, unilateral sagittal T2-weighted fast spin-echo images were obtained, and a dynamic contrast material–enhanced T1-weighted three-dimensional fast low-angle shot sequence with fat suppression (repetition time msec/ echo time msec, 4.9/1.8; matrix, 448 3 224; flip angle, 12°; field of view, 170 3 170 mm; section thickness, 1.2 mm; no gap) was performed. Between 2007 and 2009, bilateral sagittal T2weighted fast spin-echo images were

obtained, and a dynamic contrast-enhanced T1-weighted three-dimensional fast spoiled gradient-echo sequence with fat suppression (repetition time msec/echo time msec, 6.5/2.5; matrix, 256 3 160; flip angle, 10°; field of view, 200 3 200 mm; section thickness, 1.5 mm; no gap) was performed (Table 1). All symptoms, signs, and the final assessment category of each examination were recorded. If any suspicious findings were newly identified during image evaluation, the detection method (mammography, US, or MR imaging), lesion location, and level of suspicion according to the Breast Imaging Reporting and Data System were recorded, and image-guided needle localization and surgical excision were performed prior to or during curative surgery (14). MR imaging–guided biopsy was available for the evaluation of MR imaging–only visible lesions.

Histopathologic Analysis Tumor diameter, histologic type, histologic or nuclear grade, resection margin status, lymphovascular invasion, and expression of estrogen receptor, progesterone receptor, HER2, and Ki67 were evaluated on the basis of surgical histopathologic findings. For the measurement of tumor diameters, a pathologist evaluated the mastectomy specimens in serial 10-mm longitudinal sections. Grossly apparent lesions in the sections were measured, and microscopically calculated sizes for smaller lesions were included. Standardized report templates included the number and sizes of invasive components and carcinoma in situ components of the tumor. A cutoff value of 10% was used to define positivity of estrogen receptor and progesterone receptor at 103 magnification. HER2 expression was considered negative when the immunohistochemical result was negative or had staining score of 1+ and was considered positive when staining score was 3+. When fluorescence in situ hybridization was performed, the HER2 result according to fluorescence in situ hybridization was given priority over that of immunohistochemical result. The molecular

Radiology: Volume 276: Number 3—September 2015  n  radiology.rsna.org

Yi et al

subtype of the tumor was classified into luminal (hormone receptor–positive and HER2-negative), HER2-enriched (HER2-positive), and triplenegative (hormone receptor–negative and HER2-negative) subtypes (15).

Postoperative Care and Follow-up After surgery, radiation therapy, chemotherapy, or hormonal therapy was administered according to the characteristics of each patient and her tumor. All patients were examined annually with bilateral mammography and bilateral US for the surveillance of local-regional or contralateral breast recurrence. Chest radiography, chest computed tomography, bone scan, and/or whole-body fluorine 18 fluorodeoxyglucose positron emission tomography were performed for the surveillance of distant metastases. The same standard of clinical treatment was used by the same clinicians throughout the study period. Metachronous breast cancer was defined as breast cancer diagnosed after 3 months of the first primary cancer diagnosis (16). Bilateral cancers diagnosed within 3 months of the first primary cancer diagnosis were categorized as synchronous cancers and were excluded (16). The last date of data collection was March 31, 2014. Follow-up duration was calculated from the date of surgery to the last date of follow-up or to the occurrence of any event or death. DFS was defined as the time from the date of surgery to the date of first detection of cancer recurrence or to death. Patients without evidence of an event were censored on the date of the most recent follow-up (17). Recurrence site was classified as local-regional (limited to the ipsilateral breast or chest wall and/or axillary, infraclavicular, or supraclavicular lymph nodes), contralateral breast, or distant (metastasis to other parts of the body) according to the first recurrence site in each patient. Statistical Methods The Pearson x2 test or Fisher exact test for categorical variables and the 697

BREAST IMAGING: Breast Cancer Recurrence in Matched Cohort Groups

Yi et al

Table 1 MR Protocols of the Unilateral and Bilateral MR Imaging Periods Unilateral MR Imaging Fast Spin-Echo T2-weighted Fat Saturated

Fast Spin-Echo T2-weighted Fat Saturated

Three-dimensional Gradient-Echo T1-weighted Fat Saturated Series before and after Contrast Agent

Orientation Repetition time (msec) Echo time (msec) Flip angle (degree) Acquisition matrix Field of view (mm) In-plane spatial resolution (mm) Section thickness (mm) Gap Postprocessing

Sagittal 5310–9120 84–82 Not applicable 384 3 269/256 3 192 170 3 170 0.4 3 0.7/0.6 3 0.9

Sagittal 4.9 1.8 12 448 3 224 170 3 170 0.4 3 0.8

Sagittal 5500 85 Not applicable 256 3 160 200 3 200 0.8 3 1.3

Sagittal 6.5 2.5 10 256 3 160 200 3 200 0.8 3 1.3

1.2 No Not applicable

1.5 No Not applicable

Field strength Coil design

1.5 T Bilateral, dedicated breast   coil using four channels

1.2 No Subtraction, three-dimensional  maximal intensity projection, axial reformatted 1.5 T Bilateral, dedicated breast   coil using four channels

1.5 No Subtraction, three-dimensional  maximal intensity projection, axial reformatted 1.5 T Bilateral, dedicated breast   coil using eight channels

Parameter

Mann-Whitney test for continuous variables were performed to compare unmatched data. The Cox proportional hazards model was used to evaluate the effect of MR imaging on DFS outcome for unmatched data. Between the MR imaging and no MR imaging groups, a five-digit one-to-one case-control match in the propensity score by using greedy matching algorithms and logistic regression analysis was performed. Thus, each case was matched to a control case in the first five digits after the decimal point of the propensity score. Matched variables included age, histologic grade, nuclear grade, tumor size, nodal status, stage, hormone receptor status, Ki-67 status, molecular subtype, and lymphovascular invasion, which were determined to be significant variables with the Cox proportional hazards model and log-rank test (Table E2 [online]). After matching, the McNemar test and absolute standardized difference were used to assess the balance of matched variables between the MR imaging and no MR imaging groups (Table 2). The balance on distribution of propensity score in both groups before and after matching was 698

Bilateral MR Imaging

Three-dimensional Gradient-Echo T1-weighted Fat Saturated Series before and after Contrast Agent

1.5 T Bilateral, dedicated breast   coil using eight channels

also checked by using box plots (18). A marginal Cox proportional hazards model was used to analyze the independent effects of preoperative MR imaging on the local-regional, contralateral breast, and distant recurrence DFS for the matched data by using the option of COVSANDWICH in the PROC PHREG procedure, which uses a robust sandwich covariance estimate (19). PROC PHREG also contained an ID variable to identify the cluster of matched datasets so that subjects in the same pair had the same ID value. The matched pair as a discrete representation of the propensity scores was used in the ID statement of PROC PHREG. Hence, we estimated the regression parameters by using sandwich variance estimates for all marginal Cox survival models and obtained 95% profile likelihood confidence intervals (CIs) of hazard ratios (HRs). The McNemar test for categorical variables and Wilcoxon signed rank test for continuous variables were used for comparison of matched groups. DFS between the no MR imaging and MR imaging groups was compared by using Kaplan-Meier estimates. Cox proportional hazards survival analysis

and the log-rank test were performed for the unmatched data. Wald 95% CIs were obtained in the Cox proportional hazards model. All statistical analyses were performed by using software (SAS, version 9.3, SAS Institute, Cary, NC; IBM SPSS Statistics for Windows, version 20.0, IBM, Armonk, NY; R, version 2.15.0, R Foundation for Statistical Computing, Vienna, Austria). P values less than .05 were considered to indicate statistically significant differences.

Results Demographics Patients without preoperative MR imaging were one-to-one matched to patients with preoperative MR imaging so that the sample consisted of 371 patient pairs from the unilateral imaging period and 97 patient pairs from the bilateral imaging period with each pair consisting of one patient with and the other without preoperative MR imaging (Figs E1, E2 [online]). Matched characteristics of the no MR imaging and MR imaging groups in the unilateral and bilateral

radiology.rsna.org  n Radiology: Volume 276: Number 3—September 2015

Age group   ,45 years   45 years Histologic grade   I or II  III Nuclear grade   I or II  III Pathologic tumor size  category   20 mm   .20 mm Pathologic nodal status  Negative  Positive Pathologic stage   0 or I   II or III Estrogen receptor  status  Positive  Negative Progesterone receptor  status  Positive  Negative Ki-67 status   .14%   14% Molecular subtype  Luminal  HER2-enriched  Triple-negative

Variable 91 (24.5) 280 (75.5) 179 (48.2) 192 (51.8) 181 (48.8) 190 (51.2)

181 (48.8) 190 (51.2) 227 (61.2) 144 (38.8) 147 (35.6) 224 (60.4)

240 (64.7) 131 (35.3)

207 (55.8) 164 (44.2) 50 (13.5) 321 (86.5) 234 (63.1) 49 (13.2) 88 (23.7)

359 383

363 379

361 381

455 287

294 448

479 263

413 329

101 641

468 98 176

No MR Imaging Group (n = 371)*

182 560

Total No. of Patients (n = 742)

Radiology: Volume 276: Number 3—September 2015  n  radiology.rsna.org

234 (63.1) 49 (13.2) 88 (23.7)

51 (16.3) 320 (83.7)

206 (51.5) 165 (48.5)

239 (59.6) 132 (40.4)

147 (35.6) 224 (64.4)

228 (61.5) 143 (38.5)

180 (45.8) 191 (54.2)

182 (49.7) 189 (50.3)

180 (47.2) 191 (52.8)

91 (24.5) 280 (75.5)

MR Imaging Group (n = 371)*

Unilateral MR Imaging Period

.803

.317

.317

.317

..99

.317

.317

.317

.317

..99

P Value†

Matched Patient Characteristics in the Unilateral and Bilateral MR Imaging Periods

Table 2

0.001

0.008

0.005

0.006

0.000

0.006

0.005

0.005

0.005

0.000

Absolute Standardized Difference‡

108 36 50

20 174

120 94

94 100

90 104

128 66

94 100

66 128

68 126

54 140

Total No. of Patients (n = 194)

54 (55.7) 18 (18.6) 25 (25.8)

10 (10.3) 87 (89.7)

60 (61.9) 47 (48.5)

47 (48.5) 50 (51.5)

45 (46.4) 52 (53.6)

64 (66.0) 33 (34.0)

47 (48.5) 50 (51.5)

33 (34.0) 64 (66.0)

34 (35.1) 63 (64.9)

27 (27.8) 70 (72.2)

No MR Imaging (n = 97)*

54 (55.7) 18 (18.6) 25 (25.8)

10 (10.3) 87 (89.7)

60 (61.9) 47 (48.5)

47 (48.5) 50 (51.5)

45 (46.4) 52 (53.6)

64 (66.0) 33 (34.0)

47 (48.5) 50 (51.5)

33 (34.0) 64 (66.0)

34 (35.1) 63 (64.9)

27 (27.8) 70 (72.2)

MR Imaging Group (n = 97)*

Bilateral MR Imaging Period

..99

..99

..99

..99

..99

..99

..99

..99

..99

..99

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

Table 2 (continues)

P Value†

Absolute Standardized Difference‡

BREAST IMAGING: Breast Cancer Recurrence in Matched Cohort Groups Yi et al

699

700

Absolute Standardized Difference‡

0.000

P Value†

..99

P values were obtained by using the McNemar test. A P value less than .05 was considered to indicate a statistically significant difference.

Absolute standardized difference was obtained to evaluate the balance of matched variables. Absolute standardized difference of less than 0.010 was considered to indicate a balanced variable. ‡

* Data are number of patients, with percentages in parentheses.

Lymphovascular invasion  No  Yes

†

66 (68.0) 31 (32.0) .317 267 (70.1) 104 (28.3) 525 217

Variable

268 (72.2) 103 (27.8)

P Value†

0.006

132 62

66 (68.0) 31 (32.0)

MR Imaging Group (n = 97)* No MR Imaging (n = 97)* Total No. of Patients (n = 194) MR Imaging Group (n = 371)* No MR Imaging Group (n = 371)* Total No. of Patients (n = 742)

Unilateral MR Imaging Period

Matched Patient Characteristics in the Unilateral and Bilateral MR Imaging Periods

Table 2 (continued)

Absolute Standardized Difference‡

Bilateral MR Imaging Period

BREAST IMAGING: Breast Cancer Recurrence in Matched Cohort Groups

Yi et al

imaging periods are listed in Table 2. No differences were found between the two groups with respect to the matched variables. Patients were 45 years or older in 75.5% (280 of 371) of cases in the unilateral imaging period and in 72.2% (70 of 97) of cases in the bilateral imaging period for both groups. Patients with pathologic stage 0 or I constituted 35.6% (147 of 371) of patients in the unilateral imaging period and 46.4% (45 of 97) of patients in the bilateral imaging period for both groups. With regard to the type of treatment and the re-excision rate after conservation surgery, no difference was observed in the surgery type, in those who underwent or did not undergo hormonal therapy, chemotherapy, or whole-breast irradiation, as well as in the re-excision rate between the no MR imaging and MR imaging groups in the unilateral and bilateral imaging periods (Table E3 [online]).

Survival Analysis Mean follow-up was 73.7 months (range, 6.9–110.2 months) for patients from the unilateral imaging period and 65.3 months (range, 6.0–98.8 months) for patients from the bilateral imaging period. There were 97 patients (13.1%) with recurrences, including 20 localregional, 29 contralateral breast, and 48 distant recurrences, among 742 patients from the unilateral imaging period. There were 39 patients (20.1%) with recurrences, including seven local-regional, 22 contralateral breast, and 10 distant recurrences, among 194 patients from the bilateral imaging period (Table 3). Kaplan-Meier analysis of DFS in the unilateral imaging period showed that the MR imaging group had better localregional recurrence DFS than the no MR imaging group (Fig 1). There were no significant differences in total recurrence DFS, contralateral breast recurrence DFS, or distant recurrence DFS between the two groups. In the bilateral imaging period, the MR imaging group had better total recurrence DFS (Fig 2) and contralateral breast recurrence DFS than the no MR imaging group. No differences were found in local-regional recurrence DFS

or distant recurrence DFS between the two groups. Five-year contralateral breast recurrence DFS probability of the no MR imaging group was 80.4% (78 of 97) versus 99.0% (96 of 97) for the MR imaging group in the bilateral imaging period, whereas it was 96.2% (357 of 371) versus 97.8% (363 of 371) in the unilateral imaging period.

Marginal Cox Proportional Hazards Model Analysis In the unilateral imaging period, preoperative MR imaging was significantly associated with better local-regional recurrence DFS (HR, 0.33; 95% CI: 0.12, 0.91; P = .032), whereas no statistically significant association was observed for total recurrence DFS (HR, 0.80; 95% CI: 0.54, 1.19; P = .282), contralateral breast recurrence DFS (HR, 0.75; 95% CI: 0.36, 1.57; P = .440), or distant recurrence DFS (HR, 1.21; 95% CI: 0.68, 2.14; P = .515) (Table 3). In the bilateral imaging period, preoperative MR imaging was significantly associated with better total recurrence DFS. Preoperative MR imaging was associated with 85% reduction in risk of recurrence for total recurrences (HR, 0.15; 95% CI: 0.07, 0.32; P , .001), mainly from contralateral breast recurrence risk reduction (HR, 0.03; 95% CI: 0.04, 0.21; P , .001). However, no statistically significant association was observed for local-regional recurrence DFS (HR, 0.26; 95% CI: 0.03, 1.89; P = .180) or distant recurrence DFS (HR, 0.40; 95% CI: 0.11, 9.18; P = .178) (Table 3). Of the 22 metachronous contralateral breast cancers from the bilateral imaging period, only one node-negative pT1 invasive cancer (4.5%, one of 22) was detected in the MR imaging group. The remaining 21 cancers detected in the no MR imaging group varied from stage 0 to III, 38.0% (eight of 21) of which were node positive (Table 4). Discussion In this study, we compared the DFS outcomes of breast cancer patient cohorts without versus with preoperative MR imaging and subsequent curative surgery. We also performed a separate

radiology.rsna.org  n Radiology: Volume 276: Number 3—September 2015

BREAST IMAGING: Breast Cancer Recurrence in Matched Cohort Groups

Yi et al

Table 3 Marginal Cox Proportional Hazards Models of the Association between Preoperative MR Imaging and Local-Regional, Contralateral Breast, and Distant Recurrence Rates in the Unilateral and Bilateral Periods Unilateral MR Imaging Period (n = 742) Parameter Total   No MR imaging group   MR imaging group The first recurrence‡  Local-regional    No MR imaging   MR imaging   Contralateral breast    No MR imaging   MR imaging  Distant    No MR imaging   MR imaging

†

No. of Events*

HR

54 (14.6) 43 (11.6)

1.00 0.80

15 (4.0) 5 (1.4)

1.00 0.33

95% CI

Bilateral MR Imaging Period (n = 194) P Value

No. of Events*

HR†

31 (32.0) 8 (8.3)

1.00 0.15

4 (4.1) 3 (3.1)

1.00 0.26

21 (21.7) 1 (1.0)

1.00 0.03

6 (6.2) 4 (4.1)

1.00 0.40

95% CI

.282 0.54, 1.19

,.001 0.07, 0.32

.032 0.12, 0.91

.180 0.03, 1.89

.440 17 (4.6) 12 (3.2)

1.00 0.75

0.36, 1.57

,.001 0.004, 0.21

.515 22 (5.9) 26 (7.0)

1.00 1.21

0.68, 2.14

P Value

.178 0.11, 1.51

* Data in parentheses are percentages. †

A marginal Cox proportional hazards model was used to evaluate the adjusted effect of preoperative MR imaging on DFS adjusted by 11 variables listed in Table 1.

‡

Total recurrences were divided into three subgroups according to the first breast cancer recurrence in each patient.

analysis of the unilateral and bilateral imaging periods to evaluate the effect of the two different MR imaging protocols; the bilateral breast MR imaging protocol allows surveillance of the contralateral breast with high temporal resolution, whereas the unilateral breast MR imaging protocol allows refined ipsilateral breast images with high spatial resolution, forgoing surveillance of the contralateral breast as a tradeoff. Our study results showed that the metachronous contralateral breast cancer incidence was only reduced in the MR imaging group from the bilateral imaging period and not from the unilateral imaging period. Therefore, our findings support the preoperative MR imaging surveillance of the contralateral breast in these patients. Ironically, with regard to local-regional recurrence, better local-regional recurrence DFS was observed in the MR imaging group from the unilateral imaging period (HR, 0.33; 95% CI: 0.12, 0.91; P = .032) and was not consistently observed in the bilateral imaging period (HR, 0.26; 95% CI: 0.03, 1.89; P = .180). This might be explained by the relatively better spatial resolution

of the unilateral imaging protocol for the ipsilateral breast than that of the bilateral imaging protocol, leading to more accurate detection and excision of multifocal malignant foci. However, as this result was not realized from the bilateral imaging period, our results also support the conclusion of a recent meta-analysis that suggested that preoperative MR imaging does not reduce the risk of local or distant recurrence (1). However, in the meta-analysis, the researchers were not able to make a definitive conclusion regarding contralateral breast outcomes in patients with breast cancer (1). The strengths of our study were the application of two different MR imaging protocols (with and without surveillance of the contralateral breast) in consecutive patients and the usage of a marginal model for comparison of equivalent groups by controlling for 11 clinical-pathologic covariates. The reduced metachronous cancer incidence in contralateral breasts observed from the bilateral imaging period in our study might be attributed to the higher quality of the MR protocol and availability of MR imaging–guided biopsy, leading

Radiology: Volume 276: Number 3—September 2015  n  radiology.rsna.org

to the direct linkage of MR imaging interpretation and immediate histologic confirmation. This may be one of the reasons for the contradictory results from previous retrospective studies in which the contralateral breast recurrence rate was not different between the MR imaging and no MR imaging groups (11,12). Furthermore, the previous studies did not include all consecutive patients but rather included only patients who had undergone breastconserving surgery. Because candidates for breast-conserving surgery tend to have early stage cancers and have a lower probability of contralateral breast cancers, this may have caused underestimation of the role of preoperative MR imaging and may have been affected by selection bias. It has been suggested that additional cancers detected only by using MR imaging is either biologically inconsequential or can be adequately managed through adjuvant systemic therapy or hormonal therapy, which have been shown to reduce the subsequent metachronous contralateral cancer rate by 20% and 62%, respectively (20). However, according to a previous study, women who developed 701

BREAST IMAGING: Breast Cancer Recurrence in Matched Cohort Groups

Yi et al

Figure 1

Figure 1:  (a–d) Kaplan-Meier survival curves show breast cancer recurrence DFS outcomes in patients with newly diagnosed breast cancer without and with preoperative MR imaging in the unilateral imaging period for (a) total, (b) local-regional, (c) contralateral breast, and (d) distant recurrences. The MR imaging group had better local-regional recurrence DFS (b) than the no MR imaging group. There were no differences in total recurrence DFS (a), contralateral breast recurrence DFS (c), or distant recurrence DFS (d) between the two groups.

a metachronous cancer within 5 years of primary cancer detection, and who were younger than 50 years, were 3.9 times more likely to die than women with unilateral cancer. Researchers have explained these results by the dual effect of adjuvant chemotherapy on 702

metachronous cancer, as it can selectively prevent the occurrence of favorable cancers, leading to the occurrence of more aggressive cancers clinically (16). In addition, another study including 723 patients with metachronous contralateral breast cancers reported that a

time interval to second cancers of less than 3 years was a strong poor prognostic factor in distant DFS (21). So, earlier detection of contralateral breast cancers during preoperative MR imaging is expected to improve the overall survival of patients with newly diagnosed breast

radiology.rsna.org  n Radiology: Volume 276: Number 3—September 2015

BREAST IMAGING: Breast Cancer Recurrence in Matched Cohort Groups

Yi et al

Figure 2

Figure 2:  (a–d) Kaplan-Meier survival curves show breast cancer recurrence DFS outcomes in patients with newly diagnosed breast cancer without and with preoperative MR imaging in the bilateral imaging period for (a) total, (b) local-regional, (c) contralateral breast, and (d) distant recurrences. The MR imaging group had better total recurrence DFS (a) and contralateral breast recurrence DFS (c) than the no MR imaging group. No differences were found in local-regional recurrence DFS (b), or distant recurrence DFS (d) between the two groups.

cancer. Indeed, in our study, the one metachronous contralateral cancer developed in the MR imaging group from the bilateral MR imaging period was a node-negative early invasive cancer, whereas the 22 metachronous contralateral cancers developed in the no MR imaging group had more various advanced staged cancers. Our study had several limitations. First, our study cohorts were gathered from a single tertiary academic hospital

which might have differed from other institutions with regard to the patient characteristics, the direct linkage of MR imaging interpretation to surgical planning, and postoperative management. In addition, higher quality of the bilateral breast MR imaging technique and availability of MR-guided biopsy were crucial for our results. These facts may limit the generalizability of our findings. Second, we did not match for the surgery type or the type of therapy

Radiology: Volume 276: Number 3—September 2015  n  radiology.rsna.org

they received (ie, chemotherapy, radiation therapy, and hormonal therapy). However, our treatment strategy based on clinical-pathologic characteristics has not substantially changed during the study period. Although several patients who had HER2-positive cancers were enrolled in our phase II clinical trial using neoadjuvant trastuzumab, because we excluded patients who had underwent neoadjuvant chemotherapy, none of these patients were included 703

BREAST IMAGING: Breast Cancer Recurrence in Matched Cohort Groups

Yi et al

3. Peters NH, van Esser S, van den Bosch MA, et al. Preoperative MRI and surgical management in patients with nonpalpable breast cancer: the MONET - randomised controlled trial. Eur J Cancer 2011;47(6):879–886.

Table 4 Characteristics of Metachronous Contralateral Breast Recurrent Cancers in the Bilateral MR Imaging Period Bilateral MR Imaging Period (n = 194) Parameter Total no. of contralateral breast  recurrences† Pathologic stage  0  I  II  III Size of tumor (invasive)†   5 mm   6–10 mm   11–20 mm   21–50 mm   .51 mm Mean size (mm)‡ Nodal status†  Negative  Positive Mean no. of positive nodes‡

No MR Imaging Group (n = 97)

MR Imaging Group (n = 97)

P Value*

21 (21.6)

1 (1.0)

,.001 .523

3 (14.3) 6 (28.6) 7 (33.3) 5 (23.8)

0 (0) 1 (100.0) 0 (0) 0 (0)

3 (14.3) 1 (4.8) 7 (33.3) 7 (33.3) 3 (14.3) 27 (0–75)

0 (0) 1 (100.0) 0 (0) 0 (0) 0 (0) 8 (8–8)

13 (61.9) 8 (38.1) 2.7 (0–75)

1 (100.0) 0 (0) 0 (0–0)

.318

.368 .636

.368

Note.—Unless otherwise indicated, data are numbers of patients, with percentages in parentheses. * The McNemar test for categorical variables and the Wilcoxon signed rank test for continuous variables were performed. †

Limited to patients with metachronous contralateral breast cancers only (n = 22).

‡

Data in parentheses are ranges.

in this study. So, it is likely that differences in treatment had little effect on our results. Third, our dichotomous age grouping (,45 years, 45 years) does not capture the full nature of agerelated factors, including menopausal status. Fourth, there may have been selection bias as we had excluded patients who had undergone neoadjuvant chemotherapy. Fifth, the two patient populations of unilateral and bilateral MR imaging groups were derived from different periods (2004–2006 vs 2007–2009, respectively). So, although we attempted to match the patient populations by using 11 covariates, any observed differences between the two groups might have been confounded with time, such as hardware or software upgrades between the two periods. Last, our sample size was relatively small, and thus the power to detect differences in local-regional or distant recurrence was likely limited. 704

In conclusion, preoperative bilateral breast MR imaging for the staging of breast cancer reduces the risk of contralateral breast recurrence; however, no observed reduction in risk of local-regional or distant recurrence was shown. Disclosures of Conflicts of Interest: A.Y. disclosed no relevant relationships. N.C. disclosed no relevant relationships. K.S.Y. disclosed no relevant relationships. W.H. disclosed no relevant relationships. D.Y.N. disclosed no relevant relationships. W.K.M. disclosed no relevant relationships.

References 1. Houssami N, Turner R, Macaskill P, et al. An individual person data meta-analysis of preoperative magnetic resonance imaging and breast cancer recurrence. J Clin Oncol 2014;32(5):392–401. 2. Turnbull L, Brown S, Harvey I, et al. Comparative effectiveness of MRI in breast cancer (COMICE) trial: a randomised controlled trial. Lancet 2010;375(9714):563–571.

4. Houssami N, Ciatto S, Macaskill P, et al. Accuracy and surgical impact of magnetic resonance imaging in breast cancer staging: systematic review and meta-analysis in detection of multifocal and multicentric cancer. J Clin Oncol 2008;26(19):3248–3258. 5. Plana MN, Carreira C, Muriel A, et al. Magnetic resonance imaging in the preoperative assessment of patients with primary breast cancer: systematic review of diagnostic accuracy and meta-analysis. Eur Radiol 2012;22(1):26–38. 6. Early Breast Cancer Trialists’ Collaborative Group (EBCTCG), Darby S, McGale P, et al. Effect of radiotherapy after breast-conserving surgery on 10-year recurrence and 15-year breast cancer death: meta-analysis of individual patient data for 10,801 women in 17 randomised trials. Lancet 2011;378(9804):1707–1716. 7. Anderson SJ, Wapnir I, Dignam JJ, et al. Prognosis after ipsilateral breast tumor recurrence and locoregional recurrences in patients treated by breast-conserving therapy in five National Surgical Adjuvant Breast and Bowel Project protocols of node-negative breast cancer. J Clin Oncol 2009;27(15):2466–2473. 8. Pilewskie M, King TA. Magnetic resonance imaging in patients with newly diagnosed breast cancer: a review of the literature. Cancer 2014;120(14):2080–2089. 9. Fischer U, Zachariae O, Baum F, von Heyden D, Funke M, Liersch T. The influence of preoperative MRI of the breasts on recurrence rate in patients with breast cancer. Eur Radiol 2004;14(10):1725–1731. 10. Kim JY, Cho N, Koo HR, et al. Unilateral breast cancer: screening of contralateral breast by using preoperative MR imaging reduces incidence of metachronous cancer. Radiology 2013;267(1):57–66. 11. Solin LJ, Orel SG, Hwang WT, Harris EE, Schnall MD. Relationship of breast magnetic resonance imaging to outcome after breastconservation treatment with radiation for women with early-stage invasive breast carcinoma or ductal carcinoma in situ. J Clin Oncol 2008;26(3):386–391. 12. Ko ES, Han BK, Kim RB, et al. Analysis of the effect of breast magnetic resonance imaging on the outcome in women undergoing breast conservation surgery with radiation therapy. J Surg Oncol 2013;107(8):815–821.

radiology.rsna.org  n Radiology: Volume 276: Number 3—September 2015

BREAST IMAGING: Breast Cancer Recurrence in Matched Cohort Groups

13. Friedman PD, Swaminathan SV, Herman K, Kalisher L. Breast MRI: the importance of bilateral imaging. AJR Am J Roentgenol 2006;187(2):345–349.

16. Hartman M, Czene K, Reilly M, et al. Incidence and prognosis of synchronous and metachronous bilateral breast cancer. J Clin Oncol 2007;25(27):4210–4216.

14. American College of Radiology. ACR BI RADS—magnetic resonance imaging. In: ACR breast imaging reporting and data system atlas. Reston, Va: American College of Radiology, 2003.

17. Hudis CA, Barlow WE, Costantino JP, et al. Proposal for standardized definitions for efficacy end points in adjuvant breast cancer trials: the STEEP system. J Clin Oncol 2007;25(15):2127–2132.

15. Fumagalli D, Bedard PL, Nahleh Z, et al. A common language in neoadjuvant breast cancer clinical trials: proposals for standard definitions and endpoints. Lancet Oncol 2012;13(6):e240–e248.

18. Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res 2011;46(3): 399–424.

Radiology: Volume 276: Number 3—September 2015  n  radiology.rsna.org

Yi et al

19. Glidden DV, Vittinghoff E. Modelling clus tered survival data from multicentre clinical trials. Stat Med 2004;23(3):369–388. 20. Early Breast Cancer Trialists’ Collaborative Group (EBCTCG). Effects of chemotherapy and hormonal therapy for early breast cancer on recurrence and 15-year survival: an overview of the randomised trials. Lancet 2005;365(9472):1687–1717. 21. Alkner S, Bendahl PO, Fernö M, Manjer J, Rydén L. Prediction of outcome after diagnosis of metachronous contralateral breast cancer. BMC Cancer 2011;11:114–126.

705