ARTICLE IN PRESS
Original Investigation
Perfusion Parameters in Dynamic Contrast-enhanced MRI and Apparent Diffusion Coefficient Value in Diffusion-weighted MRI: Association with Prognostic Factors in Breast Cancer Hyun Sil Lee, MD, Sung Hun Kim, MD, Bong Joo Kang, MD, Ji Eun Baek, MD, Byung Joo Song, MD Rationale and Objectives: To evaluate the association of prognostic factors and subtypes of breast cancer with perfusion parameters in dynamic contrast-enhanced magnetic resonance imaging and apparent diffusion coefficient (ADC) values in diffusionweighted magnetic resonance imaging. Materials and Methods: Quantitative perfusion parameters (constant of transfer from plasma to interstitium, constant of transfer from the interstitium to the plasma, extravascular/extracellular volume per unit of volume of tissue [ve], and initial area under the concentration curve [iAUC]) and ADC values in the entire tumor volume of 52 invasive ductal carcinomas were obtained using histogram analysis. Four measures (25th percentile, mean, median, 75th percentile) were calculated for each parameter and the ADC value. Associations of perfusion parameters and ADC values with prognostic factors and tumor subtypes were analyzed. Results: Among perfusion parameters, iAUCmean and iAUCmedian were greater in tumors larger than 2 cm (8.23 ± 2.33, 8.64 ± 2.67 × 104) than in those smaller than 2 cm (6.99 ± 1.92, 7.04 ± 2.15 × 104; P = 0.046, 0.023). Ve median was higher in tumors with progesterone receptor (PR) positivity (0.54 ± 0.18) than in those with PR negativity (0.44 ± 0.1, P = 0.041). There were higher ADCmean and ADCmedian in tumors with human epidermal growth factor receptor 2 (HER2) positivity (1.306 and 1.278 × 10−3 mm2/s) than in those with HER2 negativity (1.078 and 1.053 × 10−3 mm2/s; P = 0.012 and 0.020). Higher ADCmean and ADCmedian were observed in HER2-enriched type (1.404 and 1.378 × 10−3 mm2/s) than in luminal type (1.096 and 1.073 × 10−3 mm2/s; P = 0.030 and 0.045). Conclusions: Among perfusion parameters, iAUC was associated with tumor size and ve median was associated with PR positivity. Mean and median ADC values showed positive correlation with HER2-positive and HER2-enriched tumors. Key Words: Breast cancer; magnetic resonance imaging; perfusion; ADC value; prognostic factor. © 2015 The Association of University Radiologists. Published by Elsevier Inc. All rights reserved.
INTRODUCTION
B
iopsy specimens are required for analysis of conventional prognostic factors such as tumor size, axillary lymph node status, histologic grade, and molecular marker expression (1). The availability of magnetic Acad Radiol 2016; ■:■■–■■ From the Department of Radiology, College of Medicine, Seoul St. Mary’s Hospital, The Catholic University of Korea, Banpo-daero 222, Seocho-gu, Seoul 137-701, Republic of Korea (H.S.L., S.H.K., B.J.K., J.E.B.); Department of General Surgery, College of Medicine, Seoul St. Mary’s Hospital, The Catholic University of Korea, Seoul, Republic of Korea (B.J.S.). Received July 20, 2015; revised December 15, 2015; accepted December 17, 2015. Address correspondence to: S.H.K. e-mail: [email protected] © 2015 The Association of University Radiologists. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.acra.2015.12.011
resonance imaging (MRI) has prompted efforts to develop noninvasive MRI-based biomarkers to predict the prognosis of breast cancers using techniques such as diffusion-weighted imaging (DWI) and dynamic contrast-enhanced (DCE) MRI. Various apparent diffusion coefficient (ADC) parameters derived from DWI and perfusion parameters obtained from DCEMRI have been associated with several prognostic factors (2–4), and recent studies have also reported correlations between pharmacokinetic parameters of breast DCE-MRI and prognostic factors, suggesting poorer prognosis in tumors with higher constant of transfer from plasma to interstitium (Ktrans) and constant of transfer from the interstitium to the plasma (kep) values or lower extravascular/extracellular volume per unit of volume of tissue (ve) values (3,4). However, to our knowledge, there have been no studies so far that have analyzed both 1
Academic Radiology, Vol ■, No ■, ■■ 2016
LEE ET AL
perfusion parameters and ADC values in the entire tumor volume for simultaneous correlation of prognostic factors or subtypes. The purpose of our study was to investigate the association of prognostic factors and tumor subtypes in patients diagnosed with breast cancer to both MR perfusion parameters and ADC values. In addition, we compared histogram analysis and region of interest (ROI) analysis for the prediction of prognosis of breast cancer. MATERIALS AND METHODS Patients
Institutional review board approval was obtained for this retrospective study, and informed consent was waived. Between February 2012 and March 2013, 81 consecutive breast cancer patients diagnosed by percutaneous biopsy underwent DCEMRI and DWI on a 3T MRI system for preoperative evaluation. Among them, a total of 29 patients were excluded because they had received preoperative neoadjuvant chemotherapy (n = 12), had histologic types other than invasive ductal carcinoma (IDC) (n = 11), or due to processing software failure (n = 6). Finally, a total of 52 masses from 52 patients (mean age 54.8 years, range 36–72 years) were included in the analysis of perfusion parameters and ADC values. MRI Acquisition
All breast MR examinations were performed using a Siemens MAGNETOM Verio 3.0T MRI system (Siemens Healthcare, Erlangen, Germany). The images were obtained using the following sequences: (1) an axial turbo spin-echo T2weighted imaging sequence with a repetition time (TR)/ echo time (TE) of 4530/93 ms, a flip angle of 80°, an field-ofview (FOV) of 320 × 320 mm, a matrix size of 576 × 403, a slice thickness of 4 mm, and an acquisition time of 2 min 28 s; (2) a DWI sequence (readout segmented echo planar imaging (EPI)) with two different b values (0 and 750 s/mm2), a TR/TE of 5600/55 ms, an FOV of 360 × 180 mm, a matrix size of 192 × 82, a slice thickness of 4 mm, and an acquisition time of 2 min 31 s; (3) precontrast T1-weighted three-dimensional volumetric interpolated breath-hold examinations with a TR/TE of 2.7/0.8 ms, an FOV of 320 × 320 mm, a matrix size of 256 × 192, and a slice thickness of 2 mm with different flip angles (2°, 15°); (4) DCE axial T1-weighted imaging (T1WI) with fat suppression with TR/TE of 2.5/0.8 ms, a flip angle of 10°, a slice thickness of 2.0 mm, and an acquisition time of 4 min 30 s (temporal resolution 6 s) after an intravenous bolus injection of 0.1 mmol/kg gadobutrol (Gadovist, Schering, Berlin, Germany) as recommended by the manufacturer; and (5) delayed, high-spatial resolution, contrast-enhanced axial T1WI with fat suppression, a slice thickness of 1.0 mm, and an acquisition time of 2 min 30 s. Imaging Analysis
MRI data were retrospectively reviewed and evaluated in consensus by two independent radiologists, with 10 and 4 years 2
of experience in breast MRI, who analyzed the perfusion parameters and ADC values. The radiologists knew that the patients had histopathologically confirmed IDC, but they were blinded to other clinical information such as the molecular markers or subtypes of the index tumors. Perfusion Parameters
Perfusion parameters were quantitatively analyzed using dedicated DCE-MRI software (Olea Sphere 2.3, Olea Medical, La Ciotat, France) based on extended Tofts mathematical model. Native T1 maps were generated using two different flip angles (2°, 15°). The arterial input function was obtained from the aorta or axillary artery by automatic arterial input function selection algorithm. Four perfusion parameters were used to assess tissue and vascular permeability characteristics: Ktrans (min−1), kep (min−1), ve (mL/100 mL of tissue, %), and iAUC (no unit, initial area under the concentration curve in 120 s) (5,6). For the estimation of the perfusion parameters, we semiautomatically drew a volume of interest covering the whole tumor area (Fig 1a) and manually drew a ROI at the highest Ktrans value in the automatically analyzed Ktrans-based perfusion map. The histogram analysis was performed for the entire tumor volume. Various histogram values were calculated, including 25th percentile, mean, 50th percentile (median), and 75th percentile. Only the mean value was included in ROI analysis. ADC Values
ADC values were calculated using Siemens MR OncoTreat software (Siemens Healthcare). We manually drew three ROIs on representative axial, sagittal, and coronal images of the index tumor on contrast-enhanced T1WI image (Fig 1b) and then applied the same ROIs to DWI and ADC maps of the index tumor. The software generated an entire tumor volume reconstruction, voxel-based ADC values, and a histogram of the ADC data including calculations for the 25th percentile, mean, 50th percentile (median), and 75th percentile ADC values. A ROI was manually drawn at the lowest ADC value of the index tumor on the ADC map and the mean value was evaluated. Histopathologic Analysis
Histopathologic information was obtained from pathology reports. Tumor size was regarded as the maximum diameter of the tumor on the surgical specimen. Tumors were graded according to the modified criteria of Bloom and Richardson. Estrogen receptor (ER) or progesterone receptor (PR) positivity was indicated by stained nuclei in >10% of cancer cells on 10 high-power fields. Positive Ki-67 expression was defined by a rate of Ki-67 positivity in >14% of the cancer cell nuclei (7,8). The intensity of HER2 expression was semiquantitatively scored as 0, 1+, 2+, or 3+. Tumors with
Academic Radiology, Vol ■, No ■, ■■ 2016
PERFUSION PARAMETERS IN BREAST CANCER
Figure 1. Two methods of three-dimensional semiautomatic volume segmentation in the same patient are shown on contrast-enhanced T1-weighted images (CE T1WI). (a) Commercially available software (Olea Sphere 2.3) was used for the calculation of perfusion parameters. By placing a cursor on a representative axial slice, automatic segmentation of volumes of interest (VOIs) was performed in the entire tumor based on pixel intensities. (b) CE T1WI was used on the original images to estimate apparent diffusion coefficient (ADC) values using the software (left). By drawing three regions of interest (ROIs) on representative axial, coronal, and sagittal contrast-enhanced T1WI (middle), total tumor volume was automatically reconstructed (right). This VOI was copied and applied to the ADC map for the calculation of ADC values.
a 3+ score were classified as HER2 positive and those with scores of 0 or +1 were classified as HER2 negative. For tumors with a 2+ score, gene amplification was used to determine HER2 status. Tumor subtypes were categorized by molecular marker expression as follows: luminal type (ER or PR positive, HER2 negative); HER2-enriched type (HER2 overexpressed or amplified, ER and PR negative); and triplenegative type (ER or PR negative, HER2 negative). Histopathologic assessment was performed by a pathologist (L.A.W.) with 16 years of experience. Statistical Analysis
Histopathologic prognostic factors and immunohistochemical subtypes were dichotomized and all cases were assigned to one of two groups: tumor size (T1 vs. others), axillary node metastasis (negative vs. positive), histologic grade (grades 1 and 2 vs. grade 3), ER or PR expression (≤10% vs. >10%), Ki-67 (≤14% vs. >14%), and HER-2 expression (negative vs. positive). Tumor subtypes were analyzed by paired comparison. The normality of data distribution was evaluated by the Shapiro-Wilk test. Descriptive continuous variables are presented as mean ± standard deviation and were analyzed by Student t test, Wilcoxon rank sum test,
or Kruskal-Wallis test. Pearson or Spearman correlation coefficients were calculated to correlating perfusion parameters and ADC values on histogram and ROI analysis. For some variables showing statistical significance in univariate analysis, multivariable analysis was done using analysis of covariance. All statistical analyses were performed using the SAS Enterprise Guide 5.1 software package (SAS Institute Inc, Cary, North Carolina), and P values of 2 29 0.58 ± 0.31 Lymph node metastasis Negative 32 0.57 ± 0.3 Positive 20 0.52 ± 0.2 Histologic grade Nonhigh (grades 1,2) 35 0.54 ± 0.23 High (grade 3) 17 0.57 ± 0.34 Estrogen receptor Negative 14 0.5 ± 0.16 Positive 38 0.56 ± 0.3 Progesterone receptor Negative 20 0.47 ± 0.16 Positive 32 0.59 ± 0.31 Human epidermal growth factor receptor 2 Negative 39 0.57 ± 0.28 Positive 13 0.47 ± 0.19 Ki-67 (%) cutoff 14% Negative 16 0.54 ± 0.27 Positive 36 0.55 ± 0.27 Tumor subtype Luminal 39 0.57 ± 0.29 Triple negative 9 0.5 ± 0.18 HER2 enriched 4 0.46 ± 0.14
kep Mean (min−1)
iAUC
P Value
Mean (×104)
P Value
ve Mean
P Value
P Value
0.417
0.67 ± 0.23 0.61 ± 0.19
0.461
1.25 ± 0.93 1.37 ± 0.77
0.253
11.73 ± 3.51 12.42 ± 2.95
0.261
0.955
0.65 ± 0.22 0.63 ± 0.19
0.843
1.37 ± 0.96 1.22 ± 0.61
>0.999
11.70 ± 3.47 12.78 ± 2.66
0.263
0.961
0.63 ± 0.21 0.65 ± 0.21
0.653
1.23 ± 0.65 1.5 ± 1.13
>0.999
12.29 ± 3.15 11.75 ± 3.357
0.726
0.718
0.57 ± 0.18 0.67 ± 0.22
0.134
1.38 ± 0.83 1.29 ± 0.85
0.789
12.51 ± 2.39 11.97 ± 3.46
0.599
0.232
0.55 ± 0.16 0.69 ± 0.22
0.024*
1.4 ± 0.8 1.26 ± 0.87
0.240
12.10 ± 2.19 12.13 ± 3.73
0.858
0.291
0.66 ± 0.2 0.59 ± 0.22
0.190
1.35 ± 0.86 1.22 ± 0.79
0.743
12.41 ± 2.63 11.22 ± 4.52
0.735
0.858
0.68 ± 0.21 0.62 ± 0.21
0.226
1.21 ± 0.82 1.36 ± 0.85
0.303
13.44 ± 3.04 11.52 ± 3.13
0.076
0.842
0.67 ± 0.21 0.59 ± 0.2 0.48 ± 0.05
0.206
1.31 ± 0.85 1.34 ± 0.98 1.26 ± 0.42
0.860
12.06 ± 3.47 11.73 ± 2.11 13.48 ± 2.57
0.564
trans , constant of transfer iAUC, initial area under the concentration curve; kep, constant of transfer from the interstitium to the plasma; K from plasma to interstitium; SD, standard deviation; ve, extravascular/extracellular volume per unit of volume of tissue. Data are presented as mean ± standard deviation. * Statistically significant.
4
Ktrans 25th
P Value
Ktrans Mean
P Value
Ktrans Median
P Value
0.1 ± 0.03
0.113
0.24 ± 0.11
0.126
0.23 ± 0.13
0.136
Ktrans 75th
P Value
0.38 ± 0.2
0.191
ve 25th
P Value
0.3 ± 0.11
0.071
ve Mean
P Value
0.47 ± 0.12
0.868
ve Median
P Value
0.5 ± 0.18
0.7330
ve 75th
P Value
0.63 ± 0.2
0.861
Size (cm) ≤2 >2
0.15 ± 0.1
0.3 ± 0.16
0.3 ± 0.17
0.45 ± 0.25
0.36 ± 0.1
0.47 ± 0.12
0.5 ± 0.15
0.59 ± 0.15
Lymph node metastasis Negative
0.12 ± 0.08
Positive
0.13 ± 0.08
0.670
0.28 ± 0.16
0.605
0.27 ± 0.13
0.27 ± 0.16
0.435
0.27 ± 0.14
0.42 ± 0.25
0.413
0.42 ± 0.19
0.33 ± 0.12
0.913
0.34 ± 0.1
0.48 ± 0.12
0.372
0.45 ± 0.1
0.52 ± 0.18
0.5410
0.47 ± 0.11
0.63 ± 0.19
0.560
0.58 ± 0.14
Histologic grade Nonhigh (grades 1,2)
0.11 ± 0.07
High (grade 3)
0.16 ± 0.1
0.149
0.26 ± 0.13
0.191
0.31 ± 0.17
0.25 ± 0.15
0.054
0.31 ± 0.16
0.41 ± 0.22
0.598
0.45 ± 0.26
0.32 ± 0.12
0.222
0.36 ± 0.08
0.46 ± 0.11
0.459
0.48 ± 0.12
0.5 ± 0.16
0.5520
0.51 ± 0.16
0.6 ± 0.18
Academic Radiology, Vol ■, No ■, ■■ 2016
TABLE 2. Association of Perfusion Parameters with Prognostic Factors in Histogram Analysis
0.711
0.61 ± 0.17
Estrogen receptor Negative
0.1 ± 0.05
Positive
0.14 ± 0.09
0.152
0.23 ± 0.07
0.298
0.29 ± 0.16
0.22 ± 0.07
0.190
0.29 ± 0.17
0.36 ± 0.11
0.370
0.44 ± 0.26
0.33 ± 0.1
0.726
0.34 ± 0.11
0.43 ± 0.11
0.078
0.48 ± 0.11
0.44 ± 0.11
0.0510
0.52 ± 0.17
0.54 ± 0.13
0.068
0.63 ± 0.18
Progesterone receptor 0.1 ± 0.05
Positive
0.14 ± 0.09
0.255
0.23 ± 0.06
0.288
0.3 ± 0.17
0.22 ± 0.07
0.263
0.3 ± 0.18
0.36 ± 0.11
0.323
0.46 ± 0.28
0.33 ± 0.09
0.606
0.34 ± 0.12
0.43 ± 0.1
0.054
0.49 ± 0.12
0.44 ± 0.1
0.041*
0.54 ± 0.18
0.54 ± 0.12
0.038*
0.65 ± 0.19
Human epidermal growth factor receptor 2 Negative
0.12 ± 0.08
Positive
0.13 ± 0.08
0.583
0.28 ± 0.15
0.583
0.25 ± 0.11
0.28 ± 0.16
0.849
0.25 ± 0.12
0.44 ± 0.24
0.512
0.38 ± 0.18
0.35 ± 0.11
0.156
0.3 ± 0.1
0.48 ± 0.12
0.398
0.45 ± 0.11
0.5 ± 0.13
0.3000
0.5 ± 0.23
0.62 ± 0.17
0.228
0.57 ± 0.2
Ki-67 Negative
0.11 ± 0.08
Positive
0.13 ± 0.08
0.422
0.27 ± 0.17
0.388
0.28 ± 0.13
0.27 ± 0.2
0.346
0.27 ± 0.13
0.42 ± 0.28
0.326
0.42 ± 0.21
0.32 ± 0.1
0.479
0.34 ± 0.11
0.47 ± 0.11
0.656
0.47 ± 0.12
0.51 ± 0.14
0.5720
0.5 ± 0.17
0.64 ± 0.19
0.263
0.59 ± 0.17
Subtype Luminal
0.14 ± 0.09
0.179
0.29 ± 0.16
0.246
0.29 ± 0.17
0.153
0.45 ± 0.26
0.375
0.34 ± 0.11
0.590
0.48 ± 0.11
0.098
0.52 ± 0.17
0.0700
0.63 ± 0.18
Triple negative
0.1 ± 0.05
0.23 ± 0.06
0.22 ± 0.06
0.35 ± 0.09
0.34 ± 0.12
0.46 ± 0.13
0.47 ± 0.13
0.58 ± 0.16
HER2 enriched
0.08 ± 0.01
0.2 ± 0.04
0.17 ± 0.03
0.31 ± 0.09
0.28 ± 0.05
0.37 ± 0.05
0.38 ± 0.05
0.47 ± 0.05
0.092
continued on next page
5
PERFUSION PARAMETERS IN BREAST CANCER
Negative
LEE ET AL
6 TABLE 2.
(continued).
kep 25th
P Value
≤2
0.26 ± 0.12
0.261
>2
0.35 ± 0.22
kep Mean
P Value
0.57 ± 0.31
0.302
kep Median P Value
kep 75th
P Value
0.79 ± 0.48
0.238
iAUC 25th (×104)
iAUC iAUC P Value Mean (×104) P Value Median (×104) P Value
iAUC 75th (×104)
P Value
Size (cm) 0.68 ± 0.36
0.5 ± 0.24
0.231
0.62 ± 0.32
0.93 ± 0.47
4.23 ± 1.30
0.041*
5.64 ± 2.44
6.99 ± 1.92
0.046*
8.23 ± 2.33
7.04 ± 2.15
0.023*
8.64 ± 2.67
9.70 ± 3.04
0.103
11.02 ± 2.67
Lymph node metastasis Negative
0.28 ± 0.19
Positive
0.35 ± 0.18
0.150
0.62 ± 0.38
0.232
0.65 ± 0.28
0.54 ± 0.31
0.198
0.6 ± 0.25
0.85 ± 0.54
0.211
0.9 ± 0.37
4.76 ± 1.85
0.458
5.40 ± 2.50
7.43 ± 2.19
0.297
8.10 ± 2.27
7.62 ± 2.51
0.270
8.43 ± 2.62
10.10 ± 3.08
0.296
10.97 ± 2.53
Histologic grade Nonhigh (grades 1,2)
0.28 ± 0.16
High (grade 3)
0.35 ± 0.24
0.413
0.59 ± 0.31
0.339
0.7 ± 0.4
0.54 ± 0.27
0.320
0.63 ± 0.33
0.84 ± 0.47
0.483
0.93 ± 0.49
4.83 ± 2.08
0.234
5.37 ± 2.22
7.58 ± 2.14
0.622
7.91 ± 2.45
7.70 ± 2.41
0.356
8.41 ± 2.86
10.37 ± 2.81
0.802
10.58 ± 3.13
Estrogen receptor Negative
0.26 ± 0.1
Positive
0.32 ± 0.21
0.370
0.58 ± 0.18
0.877
0.65 ± 0.39
0.53 ± 0.16
0.893
0.58 ± 0.33
0.8 ± 0.24
0.975
0.89 ± 0.54
4.99 ± 1.43
0.643
5.02 ± 2.34
7.92 ± 1.40
0.560
7.60 ± 2.47
8.25 ± 1.57
0.488
7.82 ± 2.85
10.92 ± 1.88
0.370
10.26 ± 3.18
Progesterone receptor Negative
0.28 ± 0.12
Positive
0.32 ± 0.22
0.800
0.62 ± 0.2
0.247
0.64 ± 0.41
0.54 ± 0.14
0.714
0.59 ± 0.36
0.84 ± 0.23
0.314
0.89 ± 0.58
4.96 ± 1.26
0.529
5.04 ± 2.54
7.79 ± 1.36
0.765
7.62 ± 2.65
8.15 ± 1.70
0.596
7.80 ± 2.99
10.68 ± 1.86
0.590
10.28 ± 3.40
Human epidermal growth factor receptor 2 0.32 ± 0.19 0.27 ± 0.18
0.540
0.64 ± 0.35
0.899
0.59 ± 0.33
0.59 ± 0.3
0.597
0.5 ± 0.27
0.89 ± 0.5
0.816
0.79 ± 0.41
4.99 ± 2.12
0.657
5.07 ± 2.21
7.75 ± 1.96
0.713
7.48 ± 2.97
7.95 ± 2.35
0.920
7.87 ± 3.23
10.57 ± 2.40
0.667
10.04 ± 4.12
Ki-67 Negative
0.3 ± 0.2
Positive
0.31 ± 0.18
0.684
0.63 ± 0.39
0.656
0.63 ± 0.32
0.57 ± 0.34
0.559
0.57 ± 0.27
0.89 ± 0.61
0.599
0.86 ± 0.41
4.96 ± 2.29
0.714
5.03 ± 2.07
8.06 ± 2.11
0.421
7.52 ± 2.29
8.11 ± 2.47
0.744
7.85 ± 2.63
11.13 ± 2.70
0.256
10.13 ± 2.95
Subtype Luminal
0.33 ± 0.21
Triple negative
0.23 ± 0.08
0.52 ± 0.13
0.49 ± 0.11
0.73 ± 0.17
4.65 ± 1.07
7.39 ± 0.95
7.73 ± 1.04
10.12 ± 1.41
HER2 enriched 0.27 ± 0.04
0.59 ± 0.08
0.51 ± 0.05
0.82 ± 0.15
4.81 ± 0.90
8.39 ± 1.45
8.53 ± 1.55
11.95 ± 1.9
0.304
0.66 ± 0.39
0.793
0.59 ± 0.33
0.734
0.91 ± 0.54
0.7560
5.11 ± 2.38
0.979
7.68 ± 2.49
0.763
7.92 ± 2.89
0.874
10.35 ± 3.20
0.548
trans , constant of transfer from plasma to interstitium; SD, standard deviaiAUC, initial area under the concentration curve; kep, constant of transfer from the interstitium to the plasma; K tion; ve, extravascular/extracellular volume per unit of volume of tissue. Data are presented as mean ± SD. * Statistically significant.
Academic Radiology, Vol ■, No ■, ■■ 2016
Negative Positive
n
ADC Mean (ROI) (10−3 mm2/s)
ADC 25th (VOI) (10−3 mm2/s)
P Value
ADC Mean (VOI) (10−3 mm2/s)
P Value
ADC Median (VOI) (10−3 mm2/s)
P Value
ADC 75th (VOI) (10−3 mm2/s)
P Value
0.196
0.853 ± 0.265 0.961 ± 0.266
0.101
1.066 ± 0.261 1.190 ± 0.289
0.071
1.054 ± 0.276 1.154 ± 0.292
0.156
1.273 ± 0.283 1.406 ± 0.329
0.130
0.027**
0.928 ± 0.265 0.890 ± 0.278
0.579
1.132 ± 0.272 1.142 ± 0.302
0.814
1.115 ± 0.280 1.101 ± 0.305
0.873
1.328 ± 0.294 1.377 ± 0.349
0.588
0.236
0.875 ± 0.298 0.992 ± 0.177
0.320
1.092 ± 0.303 1.225 ± 0.210
0.178
1.065 ± 0.307 1.202 ± 0.220
0.258
1.299 ± 0.331 1.445 ± 0.255
0.115
0.968
1.019 ± 0.255 0.874 ± 0.266
0.018**
1.261 ± 0.260 1.089 ± 0.278
0.011**
1.223 ± 0.270 1.068 ± 0.285
0.024**
1.482 ± 0.292 1.297 ± 0.310
0.059
0.220
1.001 ± 0.231 0.858 ± 0.279
0.031**
1.240 ± 0.252 1.070 ± 0.282
0.018**
1.212 ± 0.270 1.046 ± 0.284
0.028**
1.466 ± 0.297 1.272 ± 0.305
0.029**
0.663
0.859 ± 0.267 1.077 ± 0.206
0.011**
1.079 ± 0.276 1.306 ± 0.227
0.012**
1.053 ± 0.277 1.279 ± 0.256
0.020**
1.287 ± 0.308 1.526 ± 0.268
0.016**
0.962
0.806 ± 0.304 0.961 ± 0.240
0.140
1.014 ± 0.297 1.189 ± 0.260
0.053
0.998 ± 0.299 1.159 ± 0.271
0.102
1.210 ± 0.302 1.407 ± 0.303
0.035**
0.978
0.881 ± 0.266 0.938 ± 0.285 1.172 ± 0.097
0.032*,**
1.096 ± 0.277 1.186 ± 0.295 1.404 ± 0.113
0.030*,**
1.073 ± 0.283 1.148 ± 0.305 1.379 ± 0.131
0.045*,**
1.303 ± 0.309 1.410 ± 0.334 1.627 ± 0.162
0.115
ADC, apparent diffusion coefficient; ROI, region of interest; SD, standard deviation; VOI, volume of interest. Data are presented as mean ± SD. * Luminal vs. HER2 enriched. ** Statistically significant.
7
PERFUSION PARAMETERS IN BREAST CANCER
Size (cm) ≤2 23 0.887 ± 0.167 >2 29 0.835 ± 0.118 Lymph node metastasis Negative 32 0.893 ± 0.137 Positive 20 0.804 ± 0.138 Histologic grade Nonhigh (grades 1,2) 35 0.875 ± 0.152 High (grade 3) 17 0.824 ± 0.117 Estrogen receptor Negative 14 0.863 ± 0.147 Positive 38 0.843 ± 0.134 Progesterone receptor Negative 20 0.827 ± 0.118 Positive 32 0.878 ± 0.155 Human epidermal growth factor receptor 2 Negative 39 0.863 ± 0.147 Positive 13 0.843 ± 0.134 Ki-67 Negative 16 0.860 ± 0.173 Positive 36 0.858 ± 0.129 Subtype Luminal 39 0.859 ± 0.151 Triple negative 9 0.852 ± 0.119 HER2 enriched 4 0.871 ± 0.140
P Value
Academic Radiology, Vol ■, No ■, ■■ 2016
TABLE 3. Association of Diffusion Parameters with Prognostic Factors
Academic Radiology, Vol ■, No ■, ■■ 2016
LEE ET AL
TABLE 4. Analysis of Covariance (ANCOVA) for ADC Values with HER2, ER, and PR Status ADC 25th
P Value
Human epidermal growth factor receptor 2 Negative (n = 39) 0.887 ± 0.046 0.023* Positive (n = 13) 1.081 ± 0.071 Estrogen receptor Negative (n = 14) 1.015 ± 0.081 0.574 Positive (n = 38) 0.954 ± 0.055 Progesterone receptor Negative (n = 20) 1.021 ± 0.060 0.457 Positive (n = 32) 0.947 ± 0.071
ADC Mean
P Value
ADC Median
P Value
ADC 75th
P Value
1.114 ± 0.047 1.311 ± 0.074
0.025*
1.084 ± 0.049 1.280 ± 0.076
0.03*
1.325 ± 0.053 1.529 ± 0.083
0.038*
1.249 ± 0.084 1.176 ± 0.057
0.513
1.206 ± 0.086 1.158 ± 0.059
0.382
1.459 ± 0.094 1.394 ± 0.064
0.605
1.259 ± 0.062 1.166 ± 0.074
0.369
1.235 ± 0.064 1.129 ± 0.076
0.319
1.487 ± 0.070 1.367 ± 0.083
0.300
ADC, apparent diffusion coefficient. Values are least square means ± standard error from analyses of ANCOVA. ANCOVA model included HER2, ER, and PR. * Statistically significant.
Figure 2. 47-year-old woman presented with right invasive ductal carcinoma, luminal type. The tumor showed relatively high mean initial area under the concentration curve (iAUCmean) (10.50 × 104) and low mean apparent diffusion coefficient (ADCmean) values (0.964 × 10−3 mm2/s). Other tumor characteristics were as follows: tumor size, 2.2 cm; axillary lymph node metastasis, absent; histologic grade, 3; estrogen receptor (ER) and progesterone receptor (PR), positive; human epidermal growth factor receptor 2 (HER2), negative; Ki-67, 25%; volume of interest (VOI) mean of constant of transfer from plasma to interstitium (Ktrans), 0.851; VOI mean of extravascular/ extracellular volume per unit of volume of tissue (ve), 0.520; and VOI mean of constant of transfer from the interstitium to the plasma (k ep ), 1.773, respectively. (a) Contrast-enhanced axial T1-weighted images (T1WI) demonstrated an irregular enhancing mass in the right breast. A Ktransbased perfusion map (b) and an ADC map (c), volume segmentation on the ADC map (d), and the histogram of the ADC values (e) in the whole tumor are displayed.
type in ADC25, ADCmean, and ADCmedian, with higher ADC values in HER2-enriched type than in luminal type (Table 3; Figs 2 and 3).
correlation (Ktrans, r = 0.771; ve, r = 0.821; kep, r = 0.772; iAUC, r = 0.736) (P < 0.0001). However, the mean ADC values did not show significant correlation between histogram analysis and ROI analysis (r = 0.004, P = 0.98).
Correlation Between Histogram Analysis and ROI Analysis
DISCUSSION
Correlation of the mean of each perfusion parameter between histogram analysis and ROI analysis showed a strong positive
In this study, we evaluated the association of IDC prognostic factors and tumor subtypes with MR perfusion parameters
8
Academic Radiology, Vol ■, No ■, ■■ 2016
PERFUSION PARAMETERS IN BREAST CANCER
Figure 3. 68-year-old woman presented with right invasive ductal carcinoma, HER2enriched type. The tumor showed relatively high mean initial area under the concentration curve (iAUCmean) (9.87 × 104) and mean apparent diffusion coefficient (ADCmean) values (1.284 × 10−3 mm2/s). Other tumor characteristics were as follows: tumor size, 2.5 cm; axillary lymph node metastasis, absent; histologic grade, 2; estrogen receptor (ER) and progesterone receptor (PR), negative; human epidermal growth factor receptor 2 (HER2), positive; Ki-67, 30%; volume of interest (VOI) mean of constant of transfer from plasma to interstitium (Ktrans), 0.240; VOI mean of extravascular/ extracellular volume per unit of volume of tissue (ve), 0.446; and VOI mean of constant of transfer from the interstitium to the plasma (k ep ), 0.672, respectively. (a) Contrast-enhanced axial T1-weighted images (T1WI) demonstrated an oval enhancing mass in the right breast. A Ktransbased perfusion map (b) and an ADC map (c), volume segmentation on ADC map (d), and histogram of the ADC value (e) in the whole tumor are displayed.
and ADC values. There was an association between iAUC and tumor size in histogram analysis of perfusion parameters. Large tumor volume and high AUC have both been identified as poor prognostic factors in breast cancer (9). AUC is a measure of the amount of gadolinium contrast agent delivered to and retained by the tumor in the given time period (5) and a mixed parameter, displaying an intractable relationship with Ktrans, ve, and vp (fractional plasma volume) (10). Although the direct correlation between tumor size and AUC has not been proven, the result of the present study suggests that higher AUC values on MR imaging are poor prognostic indicators. In present study, IDC with PR and ER negativity showed lower ve values than those with positivity. Lower ve is associated with highly cellular environments characterized by more compact extravascular extracellular space, such as those that would be expected in aggressive cancers. A previous study reported that mean ve was lower in tumors with a high histologic grade and ER negativity (4), suggesting that lower ve values are also poor prognostic indicators. Tumors with ER and PR negativity are known as nonresponsive to endocrine therapies and are poor prognostic factors (1). The exact role of PR in breast cancer is not fully understood, but it is
known that the presence of PR reflects a functional ER pathway (11). Greater understanding of biochemical pathways and the role of ER and PR in cancer progression is still needed. There was no statistically significant difference between other pharmacokinetic perfusion parameters and immunohistochemical subtypes of breast cancer in histogram analysis. In previous studies, there were correlations between higher Ktrans and kep values and poor prognostic factors such as high histologic grade, Ki-67 positivity, or triple-negative subtype (3,4). These differences may be a result of differences in analysis methods including software, study population, and substantial heterogeneity in the perfusion kinetics of breast tumors. The ROI analysis of ADC values showed that ADCmean was lower in tumors with axillary lymph node metastasis. Association between lower ADC values and axillary lymph node metastasis has been reported previously (12), but other authors have reported that there was no significant difference between ADC values relating to the axillary lymph node status (13,14). In another study, ADC values for breast tumors in histogram analysis reported higher ADC values in patients with lymph node metastasis (2). These varying results may be associated with heterogeneous inclusion of histologic types of 9
LEE ET AL
breast cancer, inherent tumor heterogeneity, different analysis methods, and MR techniques including inconsistent b values for ADC mapping. In previous studies, there was no correlation between ADC values and ER or PR status (13,14), even in histogram analysis (2), but lower ADC values were observed in patients with ER or PR expression in this study. Several reports on the role of ER and PR have suggested that ER inhibits angiogenesis (15), and ER-positive tumors have high cellularity (16), while PR assists in the functional ER pathway (11). ADC values are affected by both tissue perfusion and cellularity, and ER and PR status might affect ADC values. Angiogenesis plays an important role in tumor proliferation and metastasis (17). In the present study, various ADC histogram parameters showed significantly higher values in HER2-positive tumors than in HER2-negative tumors, consistent with prior studies (2,13,14,18). In vivo, microscopic motions of water include molecular diffusion and microcirculation of blood in randomly oriented capillary networks, and these two factors affect ADC values (19). The perfusion effect makes ADC values higher than expected. Overexpression of HER2 in tumor cells is closely related with increased angiogenesis (17). However, the tumor microvasculature is leaky to circulating macromolecules and water (20), and this contributes to an increase of total extravascular extracellular fluid volume. The effect of increased microcirculation and the greater volume of extravascular extracellular fluid in HER2positive tumors and HER2-enriched subtype seemed to be associated with high ADC values. Theoretically, if HER2 is associated with angiogenesis, this would affect perfusion parameters on DCE-MRI. However, there was no significant correlation between perfusion parameters and HER2positive tumors in this study, and other studies have reported the same results (3,4). Those studies included a relatively small number of malignancies with heterogeneous tumor profiles and used different analysis methods; therefore, a prospective study with a consistent design is required in a large homogeneous tumor group. DCE-MRI parameters obtained from different analysis approaches have shown different correlations with angiogenesis markers (21). As tumor vascularity or cellularity is heterogeneous, enhancement kinetics and ADC values are dependent in part on the placement of the ROI within the tumor. Therefore, ROI analysis is inevitably operator dependent, but histogram analysis covering the whole tumor is relatively operator independent. In the present study, histogram analysis showed greater association to prognostic factors than ROI analysis, reflecting overall tumor kinetics or tumor niche. This study has several limitations. First, it was retrospectively performed. Second, the number of cases was relatively small and we did multiple comparisons using many variables. Third, although only IDC lesions were included in this study, many of which had variable amounts of ductal carcinoma in situ (DCIS) component (48 cases of 52 cases) or tumor necrosis (three cases of 52 cases), which can be potential confounding factors. Fourth, comparison of direct histopathologic 10
Academic Radiology, Vol ■, No ■, ■■ 2016
angiogenesis markers, such as microvessel density (22), vascular endothelial growth factor, and hypoxia-inducible factor 1-alpha (23) to perfusion parameters or permeability parameters, was not performed. In conclusion, among MRI perfusion parameters, significant associations were observed between iAUC and tumor size and between ve median and PR positivity in IDC of the breast. Mean and median ADC values showed positive association with tumors with HER2 positivity and HER2-enriched subtype. In addition, histogram analysis of perfusion parameters and ADC values in the entire tumor volume showed greater heterogeneity of tumor kinetics and physiology than ROI analysis. These MR perfusion or ADC histogram parameters would be applicable in breast cancer patients who received neoadjuvant chemotherapy to predict treatment response, as other parts of body such as kidney or ovary (24,25).
ACKNOWLEDGMENTS This work was based on the Olea Sphere 2.3 (Olea Medical, France) and OncoTreat (Siemens Healthcare, works in progress) postprocessing software packages. The statistical consultation was supported by a grant of the Korean Health Technology R&D Project, Ministry of Health & Welfare, Republic of Korea (HI14C1731). This study was supported by the Research Fund of Seoul St. Mary’s Hospital, The Catholic University of Korea.
REFERENCES 1. Goldhirsch A, Glick JH, Gelber RD, et al. Meeting highlights: international expert consensus on the primary therapy of early breast cancer 2005. Ann Oncol 2005; 16:1569–1583. 2. Kim EJ, Kim SH, Park GE, et al. Histogram analysis of apparent diffusion coefficient at 3.0t: correlation with prognostic factors and subtypes of invasive ductal carcinoma. J Magn Reson Imaging 2015; doi:10.1002/jmri.24934. 3. Kim JY, Kim SH, Kim YJ, et al. Enhancement parameters on dynamic contrast enhanced breast MRI: do they correlate with prognostic factors and subtypes of breast cancers? Magn Reson Imaging 2015; 33:72– 80. 4. Koo HR, Cho N, Song IC, et al. Correlation of perfusion parameters on dynamic contrast-enhanced MRI with prognostic factors and subtypes of breast cancers. J Magn Reson Imaging 2012; 36:145–151. 5. Parker GM, Buckley D. Tracer kinetic modelling for T1-weighted DCEMRI. In: Jackson A, Buckley D, Parker GM, eds. Dynamic contrastenhanced magnetic resonance imaging in oncology. Berlin Heidelberg: Springer, 2005; 81–92. 6. Tofts PS, Brix G, Buckley DL, et al. Estimating kinetic parameters from dynamic contrast-enhanced T(1)-weighted MRI of a diffusable tracer: standardized quantities and symbols. J Magn Reson Imaging 1999; 10:223– 232. 7. Cheang MC, Chia SK, Voduc D, et al. Ki67 index, HER2 status, and prognosis of patients with luminal B breast cancer. J Natl Cancer Inst 2009; 101:736–750. 8. Goldhirsch A, Winer EP, Coates AS, et al. Personalizing the treatment of women with early breast cancer: highlights of the St Gallen International Expert Consensus on the Primary Therapy of Early Breast Cancer 2013. Ann Oncol 2013; 24:2206–2223. 9. Pickles MD, Manton DJ, Lowry M, et al. Prognostic value of pretreatment DCE-MRI parameters in predicting disease free and overall survival for breast cancer patients undergoing neoadjuvant chemotherapy. Eur J Radiol 2009; 71:498–505.
Academic Radiology, Vol ■, No ■, ■■ 2016
10. Walker-Samuel S, Leach MO, Collins DJ. Evaluation of response to treatment using DCE-MRI: the relationship between initial area under the gadolinium curve (IAUGC) and quantitative pharmacokinetic analysis. Phys Med Biol 2006; 51:3593–3602. 11. Horwitz KB, McGuire WL. Specific progesterone receptors in human breast cancer. Steroids 1975; 25:497–505. 12. Razek AA, Gaballa G, Denewer A, et al. Invasive ductal carcinoma: correlation of apparent diffusion coefficient value with pathological prognostic factors. NMR Biomed 2010; 23:619–623. 13. Park SH, Choi HY, Hahn SY. Correlations between apparent diffusion coefficient values of invasive ductal carcinoma and pathologic factors on diffusion-weighted MRI at 3.0 Tesla. J Magn Reson Imaging 2015; 41:175–182. 14. Kim SH, Cha ES, Kim HS, et al. Diffusion-weighted imaging of breast cancer: correlation of the apparent diffusion coefficient value with prognostic factors. J Magn Reson Imaging 2009; 30:615–620. 15. Ludovini V, Sidoni A, Pistola L, et al. Evaluation of the prognostic role of vascular endothelial growth factor and microvessel density in stages I and II breast cancer patients. Breast Cancer Res Treat 2003; 81:159– 168. 16. Black R, Prescott R, Bers K, et al. Tumour cellularity, oestrogen receptors and prognosis in breast cancer. Clin Oncol 1983; 9:311– 318. 17. Kumar R, Yarmand-Bagheri R. The role of HER2 in angiogenesis. Semin Oncol 2001; 28(5 suppl 16):27–32.
PERFUSION PARAMETERS IN BREAST CANCER
18. Martincich L, Deantoni V, Bertotto I, et al. Correlations between diffusionweighted imaging and breast cancer biomarkers. Eur Radiol 2012; 22:1519–1528. 19. Le Bihan D, Breton E, Lallemand D, et al. Separation of diffusion and perfusion in intravoxel incoherent motion MR imaging. Radiology 1988; 168:497–505. 20. Dvorak HF, Nagy JA, Dvorak JT, et al. Identification and characterization of the blood vessels of solid tumors that are leaky to circulating macromolecules. Am J Pathol 1988; 133:95–109. 21. Su MY, Cheung YC, Fruehauf JP, et al. Correlation of dynamic contrast enhancement MRI parameters with microvessel density and VEGF for assessment of angiogenesis in breast cancer. J Magn Reson Imaging 2003; 18:467–477. 22. Weidner N. Intratumor microvessel density as a prognostic factor in cancer. Am J Pathol 1995; 147:9–19. 23. Carmeliet P, Jain RK. Angiogenesis in cancer and other diseases. Nature 2000; 407:249–257. 24. Bharwani N, Miquel ME, Powles T, et al. Diffusion-weighted and multiphase contrast-enhanced MRI as surrogate markers of response to neoadjuvant sunitinib in metastatic renal cell carcinoma. Br J Cancer 2014; 110:616– 624. 25. Kyriazi S, Collins DJ, Messiou C, et al. Metastatic ovarian and primary peritoneal cancer: assessing chemotherapy response with diffusionweighted MR imaging—value of histogram analysis of apparent diffusion coefficients. Radiology 2011; 261:182–192.
11
Original Investigation
Perfusion Parameters in Dynamic Contrast-enhanced MRI and Apparent Diffusion Coefficient Value in Diffusion-weighted MRI: Association with Prognostic Factors in Breast Cancer Hyun Sil Lee, MD, Sung Hun Kim, MD, Bong Joo Kang, MD, Ji Eun Baek, MD, Byung Joo Song, MD Rationale and Objectives: To evaluate the association of prognostic factors and subtypes of breast cancer with perfusion parameters in dynamic contrast-enhanced magnetic resonance imaging and apparent diffusion coefficient (ADC) values in diffusionweighted magnetic resonance imaging. Materials and Methods: Quantitative perfusion parameters (constant of transfer from plasma to interstitium, constant of transfer from the interstitium to the plasma, extravascular/extracellular volume per unit of volume of tissue [ve], and initial area under the concentration curve [iAUC]) and ADC values in the entire tumor volume of 52 invasive ductal carcinomas were obtained using histogram analysis. Four measures (25th percentile, mean, median, 75th percentile) were calculated for each parameter and the ADC value. Associations of perfusion parameters and ADC values with prognostic factors and tumor subtypes were analyzed. Results: Among perfusion parameters, iAUCmean and iAUCmedian were greater in tumors larger than 2 cm (8.23 ± 2.33, 8.64 ± 2.67 × 104) than in those smaller than 2 cm (6.99 ± 1.92, 7.04 ± 2.15 × 104; P = 0.046, 0.023). Ve median was higher in tumors with progesterone receptor (PR) positivity (0.54 ± 0.18) than in those with PR negativity (0.44 ± 0.1, P = 0.041). There were higher ADCmean and ADCmedian in tumors with human epidermal growth factor receptor 2 (HER2) positivity (1.306 and 1.278 × 10−3 mm2/s) than in those with HER2 negativity (1.078 and 1.053 × 10−3 mm2/s; P = 0.012 and 0.020). Higher ADCmean and ADCmedian were observed in HER2-enriched type (1.404 and 1.378 × 10−3 mm2/s) than in luminal type (1.096 and 1.073 × 10−3 mm2/s; P = 0.030 and 0.045). Conclusions: Among perfusion parameters, iAUC was associated with tumor size and ve median was associated with PR positivity. Mean and median ADC values showed positive correlation with HER2-positive and HER2-enriched tumors. Key Words: Breast cancer; magnetic resonance imaging; perfusion; ADC value; prognostic factor. © 2015 The Association of University Radiologists. Published by Elsevier Inc. All rights reserved.
INTRODUCTION
B
iopsy specimens are required for analysis of conventional prognostic factors such as tumor size, axillary lymph node status, histologic grade, and molecular marker expression (1). The availability of magnetic Acad Radiol 2016; ■:■■–■■ From the Department of Radiology, College of Medicine, Seoul St. Mary’s Hospital, The Catholic University of Korea, Banpo-daero 222, Seocho-gu, Seoul 137-701, Republic of Korea (H.S.L., S.H.K., B.J.K., J.E.B.); Department of General Surgery, College of Medicine, Seoul St. Mary’s Hospital, The Catholic University of Korea, Seoul, Republic of Korea (B.J.S.). Received July 20, 2015; revised December 15, 2015; accepted December 17, 2015. Address correspondence to: S.H.K. e-mail: [email protected] © 2015 The Association of University Radiologists. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.acra.2015.12.011
resonance imaging (MRI) has prompted efforts to develop noninvasive MRI-based biomarkers to predict the prognosis of breast cancers using techniques such as diffusion-weighted imaging (DWI) and dynamic contrast-enhanced (DCE) MRI. Various apparent diffusion coefficient (ADC) parameters derived from DWI and perfusion parameters obtained from DCEMRI have been associated with several prognostic factors (2–4), and recent studies have also reported correlations between pharmacokinetic parameters of breast DCE-MRI and prognostic factors, suggesting poorer prognosis in tumors with higher constant of transfer from plasma to interstitium (Ktrans) and constant of transfer from the interstitium to the plasma (kep) values or lower extravascular/extracellular volume per unit of volume of tissue (ve) values (3,4). However, to our knowledge, there have been no studies so far that have analyzed both 1
Academic Radiology, Vol ■, No ■, ■■ 2016
LEE ET AL
perfusion parameters and ADC values in the entire tumor volume for simultaneous correlation of prognostic factors or subtypes. The purpose of our study was to investigate the association of prognostic factors and tumor subtypes in patients diagnosed with breast cancer to both MR perfusion parameters and ADC values. In addition, we compared histogram analysis and region of interest (ROI) analysis for the prediction of prognosis of breast cancer. MATERIALS AND METHODS Patients
Institutional review board approval was obtained for this retrospective study, and informed consent was waived. Between February 2012 and March 2013, 81 consecutive breast cancer patients diagnosed by percutaneous biopsy underwent DCEMRI and DWI on a 3T MRI system for preoperative evaluation. Among them, a total of 29 patients were excluded because they had received preoperative neoadjuvant chemotherapy (n = 12), had histologic types other than invasive ductal carcinoma (IDC) (n = 11), or due to processing software failure (n = 6). Finally, a total of 52 masses from 52 patients (mean age 54.8 years, range 36–72 years) were included in the analysis of perfusion parameters and ADC values. MRI Acquisition
All breast MR examinations were performed using a Siemens MAGNETOM Verio 3.0T MRI system (Siemens Healthcare, Erlangen, Germany). The images were obtained using the following sequences: (1) an axial turbo spin-echo T2weighted imaging sequence with a repetition time (TR)/ echo time (TE) of 4530/93 ms, a flip angle of 80°, an field-ofview (FOV) of 320 × 320 mm, a matrix size of 576 × 403, a slice thickness of 4 mm, and an acquisition time of 2 min 28 s; (2) a DWI sequence (readout segmented echo planar imaging (EPI)) with two different b values (0 and 750 s/mm2), a TR/TE of 5600/55 ms, an FOV of 360 × 180 mm, a matrix size of 192 × 82, a slice thickness of 4 mm, and an acquisition time of 2 min 31 s; (3) precontrast T1-weighted three-dimensional volumetric interpolated breath-hold examinations with a TR/TE of 2.7/0.8 ms, an FOV of 320 × 320 mm, a matrix size of 256 × 192, and a slice thickness of 2 mm with different flip angles (2°, 15°); (4) DCE axial T1-weighted imaging (T1WI) with fat suppression with TR/TE of 2.5/0.8 ms, a flip angle of 10°, a slice thickness of 2.0 mm, and an acquisition time of 4 min 30 s (temporal resolution 6 s) after an intravenous bolus injection of 0.1 mmol/kg gadobutrol (Gadovist, Schering, Berlin, Germany) as recommended by the manufacturer; and (5) delayed, high-spatial resolution, contrast-enhanced axial T1WI with fat suppression, a slice thickness of 1.0 mm, and an acquisition time of 2 min 30 s. Imaging Analysis
MRI data were retrospectively reviewed and evaluated in consensus by two independent radiologists, with 10 and 4 years 2
of experience in breast MRI, who analyzed the perfusion parameters and ADC values. The radiologists knew that the patients had histopathologically confirmed IDC, but they were blinded to other clinical information such as the molecular markers or subtypes of the index tumors. Perfusion Parameters
Perfusion parameters were quantitatively analyzed using dedicated DCE-MRI software (Olea Sphere 2.3, Olea Medical, La Ciotat, France) based on extended Tofts mathematical model. Native T1 maps were generated using two different flip angles (2°, 15°). The arterial input function was obtained from the aorta or axillary artery by automatic arterial input function selection algorithm. Four perfusion parameters were used to assess tissue and vascular permeability characteristics: Ktrans (min−1), kep (min−1), ve (mL/100 mL of tissue, %), and iAUC (no unit, initial area under the concentration curve in 120 s) (5,6). For the estimation of the perfusion parameters, we semiautomatically drew a volume of interest covering the whole tumor area (Fig 1a) and manually drew a ROI at the highest Ktrans value in the automatically analyzed Ktrans-based perfusion map. The histogram analysis was performed for the entire tumor volume. Various histogram values were calculated, including 25th percentile, mean, 50th percentile (median), and 75th percentile. Only the mean value was included in ROI analysis. ADC Values
ADC values were calculated using Siemens MR OncoTreat software (Siemens Healthcare). We manually drew three ROIs on representative axial, sagittal, and coronal images of the index tumor on contrast-enhanced T1WI image (Fig 1b) and then applied the same ROIs to DWI and ADC maps of the index tumor. The software generated an entire tumor volume reconstruction, voxel-based ADC values, and a histogram of the ADC data including calculations for the 25th percentile, mean, 50th percentile (median), and 75th percentile ADC values. A ROI was manually drawn at the lowest ADC value of the index tumor on the ADC map and the mean value was evaluated. Histopathologic Analysis
Histopathologic information was obtained from pathology reports. Tumor size was regarded as the maximum diameter of the tumor on the surgical specimen. Tumors were graded according to the modified criteria of Bloom and Richardson. Estrogen receptor (ER) or progesterone receptor (PR) positivity was indicated by stained nuclei in >10% of cancer cells on 10 high-power fields. Positive Ki-67 expression was defined by a rate of Ki-67 positivity in >14% of the cancer cell nuclei (7,8). The intensity of HER2 expression was semiquantitatively scored as 0, 1+, 2+, or 3+. Tumors with
Academic Radiology, Vol ■, No ■, ■■ 2016
PERFUSION PARAMETERS IN BREAST CANCER
Figure 1. Two methods of three-dimensional semiautomatic volume segmentation in the same patient are shown on contrast-enhanced T1-weighted images (CE T1WI). (a) Commercially available software (Olea Sphere 2.3) was used for the calculation of perfusion parameters. By placing a cursor on a representative axial slice, automatic segmentation of volumes of interest (VOIs) was performed in the entire tumor based on pixel intensities. (b) CE T1WI was used on the original images to estimate apparent diffusion coefficient (ADC) values using the software (left). By drawing three regions of interest (ROIs) on representative axial, coronal, and sagittal contrast-enhanced T1WI (middle), total tumor volume was automatically reconstructed (right). This VOI was copied and applied to the ADC map for the calculation of ADC values.
a 3+ score were classified as HER2 positive and those with scores of 0 or +1 were classified as HER2 negative. For tumors with a 2+ score, gene amplification was used to determine HER2 status. Tumor subtypes were categorized by molecular marker expression as follows: luminal type (ER or PR positive, HER2 negative); HER2-enriched type (HER2 overexpressed or amplified, ER and PR negative); and triplenegative type (ER or PR negative, HER2 negative). Histopathologic assessment was performed by a pathologist (L.A.W.) with 16 years of experience. Statistical Analysis
Histopathologic prognostic factors and immunohistochemical subtypes were dichotomized and all cases were assigned to one of two groups: tumor size (T1 vs. others), axillary node metastasis (negative vs. positive), histologic grade (grades 1 and 2 vs. grade 3), ER or PR expression (≤10% vs. >10%), Ki-67 (≤14% vs. >14%), and HER-2 expression (negative vs. positive). Tumor subtypes were analyzed by paired comparison. The normality of data distribution was evaluated by the Shapiro-Wilk test. Descriptive continuous variables are presented as mean ± standard deviation and were analyzed by Student t test, Wilcoxon rank sum test,
or Kruskal-Wallis test. Pearson or Spearman correlation coefficients were calculated to correlating perfusion parameters and ADC values on histogram and ROI analysis. For some variables showing statistical significance in univariate analysis, multivariable analysis was done using analysis of covariance. All statistical analyses were performed using the SAS Enterprise Guide 5.1 software package (SAS Institute Inc, Cary, North Carolina), and P values of 2 29 0.58 ± 0.31 Lymph node metastasis Negative 32 0.57 ± 0.3 Positive 20 0.52 ± 0.2 Histologic grade Nonhigh (grades 1,2) 35 0.54 ± 0.23 High (grade 3) 17 0.57 ± 0.34 Estrogen receptor Negative 14 0.5 ± 0.16 Positive 38 0.56 ± 0.3 Progesterone receptor Negative 20 0.47 ± 0.16 Positive 32 0.59 ± 0.31 Human epidermal growth factor receptor 2 Negative 39 0.57 ± 0.28 Positive 13 0.47 ± 0.19 Ki-67 (%) cutoff 14% Negative 16 0.54 ± 0.27 Positive 36 0.55 ± 0.27 Tumor subtype Luminal 39 0.57 ± 0.29 Triple negative 9 0.5 ± 0.18 HER2 enriched 4 0.46 ± 0.14
kep Mean (min−1)
iAUC
P Value
Mean (×104)
P Value
ve Mean
P Value
P Value
0.417
0.67 ± 0.23 0.61 ± 0.19
0.461
1.25 ± 0.93 1.37 ± 0.77
0.253
11.73 ± 3.51 12.42 ± 2.95
0.261
0.955
0.65 ± 0.22 0.63 ± 0.19
0.843
1.37 ± 0.96 1.22 ± 0.61
>0.999
11.70 ± 3.47 12.78 ± 2.66
0.263
0.961
0.63 ± 0.21 0.65 ± 0.21
0.653
1.23 ± 0.65 1.5 ± 1.13
>0.999
12.29 ± 3.15 11.75 ± 3.357
0.726
0.718
0.57 ± 0.18 0.67 ± 0.22
0.134
1.38 ± 0.83 1.29 ± 0.85
0.789
12.51 ± 2.39 11.97 ± 3.46
0.599
0.232
0.55 ± 0.16 0.69 ± 0.22
0.024*
1.4 ± 0.8 1.26 ± 0.87
0.240
12.10 ± 2.19 12.13 ± 3.73
0.858
0.291
0.66 ± 0.2 0.59 ± 0.22
0.190
1.35 ± 0.86 1.22 ± 0.79
0.743
12.41 ± 2.63 11.22 ± 4.52
0.735
0.858
0.68 ± 0.21 0.62 ± 0.21
0.226
1.21 ± 0.82 1.36 ± 0.85
0.303
13.44 ± 3.04 11.52 ± 3.13
0.076
0.842
0.67 ± 0.21 0.59 ± 0.2 0.48 ± 0.05
0.206
1.31 ± 0.85 1.34 ± 0.98 1.26 ± 0.42
0.860
12.06 ± 3.47 11.73 ± 2.11 13.48 ± 2.57
0.564
trans , constant of transfer iAUC, initial area under the concentration curve; kep, constant of transfer from the interstitium to the plasma; K from plasma to interstitium; SD, standard deviation; ve, extravascular/extracellular volume per unit of volume of tissue. Data are presented as mean ± standard deviation. * Statistically significant.
4
Ktrans 25th
P Value
Ktrans Mean
P Value
Ktrans Median
P Value
0.1 ± 0.03
0.113
0.24 ± 0.11
0.126
0.23 ± 0.13
0.136
Ktrans 75th
P Value
0.38 ± 0.2
0.191
ve 25th
P Value
0.3 ± 0.11
0.071
ve Mean
P Value
0.47 ± 0.12
0.868
ve Median
P Value
0.5 ± 0.18
0.7330
ve 75th
P Value
0.63 ± 0.2
0.861
Size (cm) ≤2 >2
0.15 ± 0.1
0.3 ± 0.16
0.3 ± 0.17
0.45 ± 0.25
0.36 ± 0.1
0.47 ± 0.12
0.5 ± 0.15
0.59 ± 0.15
Lymph node metastasis Negative
0.12 ± 0.08
Positive
0.13 ± 0.08
0.670
0.28 ± 0.16
0.605
0.27 ± 0.13
0.27 ± 0.16
0.435
0.27 ± 0.14
0.42 ± 0.25
0.413
0.42 ± 0.19
0.33 ± 0.12
0.913
0.34 ± 0.1
0.48 ± 0.12
0.372
0.45 ± 0.1
0.52 ± 0.18
0.5410
0.47 ± 0.11
0.63 ± 0.19
0.560
0.58 ± 0.14
Histologic grade Nonhigh (grades 1,2)
0.11 ± 0.07
High (grade 3)
0.16 ± 0.1
0.149
0.26 ± 0.13
0.191
0.31 ± 0.17
0.25 ± 0.15
0.054
0.31 ± 0.16
0.41 ± 0.22
0.598
0.45 ± 0.26
0.32 ± 0.12
0.222
0.36 ± 0.08
0.46 ± 0.11
0.459
0.48 ± 0.12
0.5 ± 0.16
0.5520
0.51 ± 0.16
0.6 ± 0.18
Academic Radiology, Vol ■, No ■, ■■ 2016
TABLE 2. Association of Perfusion Parameters with Prognostic Factors in Histogram Analysis
0.711
0.61 ± 0.17
Estrogen receptor Negative
0.1 ± 0.05
Positive
0.14 ± 0.09
0.152
0.23 ± 0.07
0.298
0.29 ± 0.16
0.22 ± 0.07
0.190
0.29 ± 0.17
0.36 ± 0.11
0.370
0.44 ± 0.26
0.33 ± 0.1
0.726
0.34 ± 0.11
0.43 ± 0.11
0.078
0.48 ± 0.11
0.44 ± 0.11
0.0510
0.52 ± 0.17
0.54 ± 0.13
0.068
0.63 ± 0.18
Progesterone receptor 0.1 ± 0.05
Positive
0.14 ± 0.09
0.255
0.23 ± 0.06
0.288
0.3 ± 0.17
0.22 ± 0.07
0.263
0.3 ± 0.18
0.36 ± 0.11
0.323
0.46 ± 0.28
0.33 ± 0.09
0.606
0.34 ± 0.12
0.43 ± 0.1
0.054
0.49 ± 0.12
0.44 ± 0.1
0.041*
0.54 ± 0.18
0.54 ± 0.12
0.038*
0.65 ± 0.19
Human epidermal growth factor receptor 2 Negative
0.12 ± 0.08
Positive
0.13 ± 0.08
0.583
0.28 ± 0.15
0.583
0.25 ± 0.11
0.28 ± 0.16
0.849
0.25 ± 0.12
0.44 ± 0.24
0.512
0.38 ± 0.18
0.35 ± 0.11
0.156
0.3 ± 0.1
0.48 ± 0.12
0.398
0.45 ± 0.11
0.5 ± 0.13
0.3000
0.5 ± 0.23
0.62 ± 0.17
0.228
0.57 ± 0.2
Ki-67 Negative
0.11 ± 0.08
Positive
0.13 ± 0.08
0.422
0.27 ± 0.17
0.388
0.28 ± 0.13
0.27 ± 0.2
0.346
0.27 ± 0.13
0.42 ± 0.28
0.326
0.42 ± 0.21
0.32 ± 0.1
0.479
0.34 ± 0.11
0.47 ± 0.11
0.656
0.47 ± 0.12
0.51 ± 0.14
0.5720
0.5 ± 0.17
0.64 ± 0.19
0.263
0.59 ± 0.17
Subtype Luminal
0.14 ± 0.09
0.179
0.29 ± 0.16
0.246
0.29 ± 0.17
0.153
0.45 ± 0.26
0.375
0.34 ± 0.11
0.590
0.48 ± 0.11
0.098
0.52 ± 0.17
0.0700
0.63 ± 0.18
Triple negative
0.1 ± 0.05
0.23 ± 0.06
0.22 ± 0.06
0.35 ± 0.09
0.34 ± 0.12
0.46 ± 0.13
0.47 ± 0.13
0.58 ± 0.16
HER2 enriched
0.08 ± 0.01
0.2 ± 0.04
0.17 ± 0.03
0.31 ± 0.09
0.28 ± 0.05
0.37 ± 0.05
0.38 ± 0.05
0.47 ± 0.05
0.092
continued on next page
5
PERFUSION PARAMETERS IN BREAST CANCER
Negative
LEE ET AL
6 TABLE 2.
(continued).
kep 25th
P Value
≤2
0.26 ± 0.12
0.261
>2
0.35 ± 0.22
kep Mean
P Value
0.57 ± 0.31
0.302
kep Median P Value
kep 75th
P Value
0.79 ± 0.48
0.238
iAUC 25th (×104)
iAUC iAUC P Value Mean (×104) P Value Median (×104) P Value
iAUC 75th (×104)
P Value
Size (cm) 0.68 ± 0.36
0.5 ± 0.24
0.231
0.62 ± 0.32
0.93 ± 0.47
4.23 ± 1.30
0.041*
5.64 ± 2.44
6.99 ± 1.92
0.046*
8.23 ± 2.33
7.04 ± 2.15
0.023*
8.64 ± 2.67
9.70 ± 3.04
0.103
11.02 ± 2.67
Lymph node metastasis Negative
0.28 ± 0.19
Positive
0.35 ± 0.18
0.150
0.62 ± 0.38
0.232
0.65 ± 0.28
0.54 ± 0.31
0.198
0.6 ± 0.25
0.85 ± 0.54
0.211
0.9 ± 0.37
4.76 ± 1.85
0.458
5.40 ± 2.50
7.43 ± 2.19
0.297
8.10 ± 2.27
7.62 ± 2.51
0.270
8.43 ± 2.62
10.10 ± 3.08
0.296
10.97 ± 2.53
Histologic grade Nonhigh (grades 1,2)
0.28 ± 0.16
High (grade 3)
0.35 ± 0.24
0.413
0.59 ± 0.31
0.339
0.7 ± 0.4
0.54 ± 0.27
0.320
0.63 ± 0.33
0.84 ± 0.47
0.483
0.93 ± 0.49
4.83 ± 2.08
0.234
5.37 ± 2.22
7.58 ± 2.14
0.622
7.91 ± 2.45
7.70 ± 2.41
0.356
8.41 ± 2.86
10.37 ± 2.81
0.802
10.58 ± 3.13
Estrogen receptor Negative
0.26 ± 0.1
Positive
0.32 ± 0.21
0.370
0.58 ± 0.18
0.877
0.65 ± 0.39
0.53 ± 0.16
0.893
0.58 ± 0.33
0.8 ± 0.24
0.975
0.89 ± 0.54
4.99 ± 1.43
0.643
5.02 ± 2.34
7.92 ± 1.40
0.560
7.60 ± 2.47
8.25 ± 1.57
0.488
7.82 ± 2.85
10.92 ± 1.88
0.370
10.26 ± 3.18
Progesterone receptor Negative
0.28 ± 0.12
Positive
0.32 ± 0.22
0.800
0.62 ± 0.2
0.247
0.64 ± 0.41
0.54 ± 0.14
0.714
0.59 ± 0.36
0.84 ± 0.23
0.314
0.89 ± 0.58
4.96 ± 1.26
0.529
5.04 ± 2.54
7.79 ± 1.36
0.765
7.62 ± 2.65
8.15 ± 1.70
0.596
7.80 ± 2.99
10.68 ± 1.86
0.590
10.28 ± 3.40
Human epidermal growth factor receptor 2 0.32 ± 0.19 0.27 ± 0.18
0.540
0.64 ± 0.35
0.899
0.59 ± 0.33
0.59 ± 0.3
0.597
0.5 ± 0.27
0.89 ± 0.5
0.816
0.79 ± 0.41
4.99 ± 2.12
0.657
5.07 ± 2.21
7.75 ± 1.96
0.713
7.48 ± 2.97
7.95 ± 2.35
0.920
7.87 ± 3.23
10.57 ± 2.40
0.667
10.04 ± 4.12
Ki-67 Negative
0.3 ± 0.2
Positive
0.31 ± 0.18
0.684
0.63 ± 0.39
0.656
0.63 ± 0.32
0.57 ± 0.34
0.559
0.57 ± 0.27
0.89 ± 0.61
0.599
0.86 ± 0.41
4.96 ± 2.29
0.714
5.03 ± 2.07
8.06 ± 2.11
0.421
7.52 ± 2.29
8.11 ± 2.47
0.744
7.85 ± 2.63
11.13 ± 2.70
0.256
10.13 ± 2.95
Subtype Luminal
0.33 ± 0.21
Triple negative
0.23 ± 0.08
0.52 ± 0.13
0.49 ± 0.11
0.73 ± 0.17
4.65 ± 1.07
7.39 ± 0.95
7.73 ± 1.04
10.12 ± 1.41
HER2 enriched 0.27 ± 0.04
0.59 ± 0.08
0.51 ± 0.05
0.82 ± 0.15
4.81 ± 0.90
8.39 ± 1.45
8.53 ± 1.55
11.95 ± 1.9
0.304
0.66 ± 0.39
0.793
0.59 ± 0.33
0.734
0.91 ± 0.54
0.7560
5.11 ± 2.38
0.979
7.68 ± 2.49
0.763
7.92 ± 2.89
0.874
10.35 ± 3.20
0.548
trans , constant of transfer from plasma to interstitium; SD, standard deviaiAUC, initial area under the concentration curve; kep, constant of transfer from the interstitium to the plasma; K tion; ve, extravascular/extracellular volume per unit of volume of tissue. Data are presented as mean ± SD. * Statistically significant.
Academic Radiology, Vol ■, No ■, ■■ 2016
Negative Positive
n
ADC Mean (ROI) (10−3 mm2/s)
ADC 25th (VOI) (10−3 mm2/s)
P Value
ADC Mean (VOI) (10−3 mm2/s)
P Value
ADC Median (VOI) (10−3 mm2/s)
P Value
ADC 75th (VOI) (10−3 mm2/s)
P Value
0.196
0.853 ± 0.265 0.961 ± 0.266
0.101
1.066 ± 0.261 1.190 ± 0.289
0.071
1.054 ± 0.276 1.154 ± 0.292
0.156
1.273 ± 0.283 1.406 ± 0.329
0.130
0.027**
0.928 ± 0.265 0.890 ± 0.278
0.579
1.132 ± 0.272 1.142 ± 0.302
0.814
1.115 ± 0.280 1.101 ± 0.305
0.873
1.328 ± 0.294 1.377 ± 0.349
0.588
0.236
0.875 ± 0.298 0.992 ± 0.177
0.320
1.092 ± 0.303 1.225 ± 0.210
0.178
1.065 ± 0.307 1.202 ± 0.220
0.258
1.299 ± 0.331 1.445 ± 0.255
0.115
0.968
1.019 ± 0.255 0.874 ± 0.266
0.018**
1.261 ± 0.260 1.089 ± 0.278
0.011**
1.223 ± 0.270 1.068 ± 0.285
0.024**
1.482 ± 0.292 1.297 ± 0.310
0.059
0.220
1.001 ± 0.231 0.858 ± 0.279
0.031**
1.240 ± 0.252 1.070 ± 0.282
0.018**
1.212 ± 0.270 1.046 ± 0.284
0.028**
1.466 ± 0.297 1.272 ± 0.305
0.029**
0.663
0.859 ± 0.267 1.077 ± 0.206
0.011**
1.079 ± 0.276 1.306 ± 0.227
0.012**
1.053 ± 0.277 1.279 ± 0.256
0.020**
1.287 ± 0.308 1.526 ± 0.268
0.016**
0.962
0.806 ± 0.304 0.961 ± 0.240
0.140
1.014 ± 0.297 1.189 ± 0.260
0.053
0.998 ± 0.299 1.159 ± 0.271
0.102
1.210 ± 0.302 1.407 ± 0.303
0.035**
0.978
0.881 ± 0.266 0.938 ± 0.285 1.172 ± 0.097
0.032*,**
1.096 ± 0.277 1.186 ± 0.295 1.404 ± 0.113
0.030*,**
1.073 ± 0.283 1.148 ± 0.305 1.379 ± 0.131
0.045*,**
1.303 ± 0.309 1.410 ± 0.334 1.627 ± 0.162
0.115
ADC, apparent diffusion coefficient; ROI, region of interest; SD, standard deviation; VOI, volume of interest. Data are presented as mean ± SD. * Luminal vs. HER2 enriched. ** Statistically significant.
7
PERFUSION PARAMETERS IN BREAST CANCER
Size (cm) ≤2 23 0.887 ± 0.167 >2 29 0.835 ± 0.118 Lymph node metastasis Negative 32 0.893 ± 0.137 Positive 20 0.804 ± 0.138 Histologic grade Nonhigh (grades 1,2) 35 0.875 ± 0.152 High (grade 3) 17 0.824 ± 0.117 Estrogen receptor Negative 14 0.863 ± 0.147 Positive 38 0.843 ± 0.134 Progesterone receptor Negative 20 0.827 ± 0.118 Positive 32 0.878 ± 0.155 Human epidermal growth factor receptor 2 Negative 39 0.863 ± 0.147 Positive 13 0.843 ± 0.134 Ki-67 Negative 16 0.860 ± 0.173 Positive 36 0.858 ± 0.129 Subtype Luminal 39 0.859 ± 0.151 Triple negative 9 0.852 ± 0.119 HER2 enriched 4 0.871 ± 0.140
P Value
Academic Radiology, Vol ■, No ■, ■■ 2016
TABLE 3. Association of Diffusion Parameters with Prognostic Factors
Academic Radiology, Vol ■, No ■, ■■ 2016
LEE ET AL
TABLE 4. Analysis of Covariance (ANCOVA) for ADC Values with HER2, ER, and PR Status ADC 25th
P Value
Human epidermal growth factor receptor 2 Negative (n = 39) 0.887 ± 0.046 0.023* Positive (n = 13) 1.081 ± 0.071 Estrogen receptor Negative (n = 14) 1.015 ± 0.081 0.574 Positive (n = 38) 0.954 ± 0.055 Progesterone receptor Negative (n = 20) 1.021 ± 0.060 0.457 Positive (n = 32) 0.947 ± 0.071
ADC Mean
P Value
ADC Median
P Value
ADC 75th
P Value
1.114 ± 0.047 1.311 ± 0.074
0.025*
1.084 ± 0.049 1.280 ± 0.076
0.03*
1.325 ± 0.053 1.529 ± 0.083
0.038*
1.249 ± 0.084 1.176 ± 0.057
0.513
1.206 ± 0.086 1.158 ± 0.059
0.382
1.459 ± 0.094 1.394 ± 0.064
0.605
1.259 ± 0.062 1.166 ± 0.074
0.369
1.235 ± 0.064 1.129 ± 0.076
0.319
1.487 ± 0.070 1.367 ± 0.083
0.300
ADC, apparent diffusion coefficient. Values are least square means ± standard error from analyses of ANCOVA. ANCOVA model included HER2, ER, and PR. * Statistically significant.
Figure 2. 47-year-old woman presented with right invasive ductal carcinoma, luminal type. The tumor showed relatively high mean initial area under the concentration curve (iAUCmean) (10.50 × 104) and low mean apparent diffusion coefficient (ADCmean) values (0.964 × 10−3 mm2/s). Other tumor characteristics were as follows: tumor size, 2.2 cm; axillary lymph node metastasis, absent; histologic grade, 3; estrogen receptor (ER) and progesterone receptor (PR), positive; human epidermal growth factor receptor 2 (HER2), negative; Ki-67, 25%; volume of interest (VOI) mean of constant of transfer from plasma to interstitium (Ktrans), 0.851; VOI mean of extravascular/ extracellular volume per unit of volume of tissue (ve), 0.520; and VOI mean of constant of transfer from the interstitium to the plasma (k ep ), 1.773, respectively. (a) Contrast-enhanced axial T1-weighted images (T1WI) demonstrated an irregular enhancing mass in the right breast. A Ktransbased perfusion map (b) and an ADC map (c), volume segmentation on the ADC map (d), and the histogram of the ADC values (e) in the whole tumor are displayed.
type in ADC25, ADCmean, and ADCmedian, with higher ADC values in HER2-enriched type than in luminal type (Table 3; Figs 2 and 3).
correlation (Ktrans, r = 0.771; ve, r = 0.821; kep, r = 0.772; iAUC, r = 0.736) (P < 0.0001). However, the mean ADC values did not show significant correlation between histogram analysis and ROI analysis (r = 0.004, P = 0.98).
Correlation Between Histogram Analysis and ROI Analysis
DISCUSSION
Correlation of the mean of each perfusion parameter between histogram analysis and ROI analysis showed a strong positive
In this study, we evaluated the association of IDC prognostic factors and tumor subtypes with MR perfusion parameters
8
Academic Radiology, Vol ■, No ■, ■■ 2016
PERFUSION PARAMETERS IN BREAST CANCER
Figure 3. 68-year-old woman presented with right invasive ductal carcinoma, HER2enriched type. The tumor showed relatively high mean initial area under the concentration curve (iAUCmean) (9.87 × 104) and mean apparent diffusion coefficient (ADCmean) values (1.284 × 10−3 mm2/s). Other tumor characteristics were as follows: tumor size, 2.5 cm; axillary lymph node metastasis, absent; histologic grade, 2; estrogen receptor (ER) and progesterone receptor (PR), negative; human epidermal growth factor receptor 2 (HER2), positive; Ki-67, 30%; volume of interest (VOI) mean of constant of transfer from plasma to interstitium (Ktrans), 0.240; VOI mean of extravascular/ extracellular volume per unit of volume of tissue (ve), 0.446; and VOI mean of constant of transfer from the interstitium to the plasma (k ep ), 0.672, respectively. (a) Contrast-enhanced axial T1-weighted images (T1WI) demonstrated an oval enhancing mass in the right breast. A Ktransbased perfusion map (b) and an ADC map (c), volume segmentation on ADC map (d), and histogram of the ADC value (e) in the whole tumor are displayed.
and ADC values. There was an association between iAUC and tumor size in histogram analysis of perfusion parameters. Large tumor volume and high AUC have both been identified as poor prognostic factors in breast cancer (9). AUC is a measure of the amount of gadolinium contrast agent delivered to and retained by the tumor in the given time period (5) and a mixed parameter, displaying an intractable relationship with Ktrans, ve, and vp (fractional plasma volume) (10). Although the direct correlation between tumor size and AUC has not been proven, the result of the present study suggests that higher AUC values on MR imaging are poor prognostic indicators. In present study, IDC with PR and ER negativity showed lower ve values than those with positivity. Lower ve is associated with highly cellular environments characterized by more compact extravascular extracellular space, such as those that would be expected in aggressive cancers. A previous study reported that mean ve was lower in tumors with a high histologic grade and ER negativity (4), suggesting that lower ve values are also poor prognostic indicators. Tumors with ER and PR negativity are known as nonresponsive to endocrine therapies and are poor prognostic factors (1). The exact role of PR in breast cancer is not fully understood, but it is
known that the presence of PR reflects a functional ER pathway (11). Greater understanding of biochemical pathways and the role of ER and PR in cancer progression is still needed. There was no statistically significant difference between other pharmacokinetic perfusion parameters and immunohistochemical subtypes of breast cancer in histogram analysis. In previous studies, there were correlations between higher Ktrans and kep values and poor prognostic factors such as high histologic grade, Ki-67 positivity, or triple-negative subtype (3,4). These differences may be a result of differences in analysis methods including software, study population, and substantial heterogeneity in the perfusion kinetics of breast tumors. The ROI analysis of ADC values showed that ADCmean was lower in tumors with axillary lymph node metastasis. Association between lower ADC values and axillary lymph node metastasis has been reported previously (12), but other authors have reported that there was no significant difference between ADC values relating to the axillary lymph node status (13,14). In another study, ADC values for breast tumors in histogram analysis reported higher ADC values in patients with lymph node metastasis (2). These varying results may be associated with heterogeneous inclusion of histologic types of 9
LEE ET AL
breast cancer, inherent tumor heterogeneity, different analysis methods, and MR techniques including inconsistent b values for ADC mapping. In previous studies, there was no correlation between ADC values and ER or PR status (13,14), even in histogram analysis (2), but lower ADC values were observed in patients with ER or PR expression in this study. Several reports on the role of ER and PR have suggested that ER inhibits angiogenesis (15), and ER-positive tumors have high cellularity (16), while PR assists in the functional ER pathway (11). ADC values are affected by both tissue perfusion and cellularity, and ER and PR status might affect ADC values. Angiogenesis plays an important role in tumor proliferation and metastasis (17). In the present study, various ADC histogram parameters showed significantly higher values in HER2-positive tumors than in HER2-negative tumors, consistent with prior studies (2,13,14,18). In vivo, microscopic motions of water include molecular diffusion and microcirculation of blood in randomly oriented capillary networks, and these two factors affect ADC values (19). The perfusion effect makes ADC values higher than expected. Overexpression of HER2 in tumor cells is closely related with increased angiogenesis (17). However, the tumor microvasculature is leaky to circulating macromolecules and water (20), and this contributes to an increase of total extravascular extracellular fluid volume. The effect of increased microcirculation and the greater volume of extravascular extracellular fluid in HER2positive tumors and HER2-enriched subtype seemed to be associated with high ADC values. Theoretically, if HER2 is associated with angiogenesis, this would affect perfusion parameters on DCE-MRI. However, there was no significant correlation between perfusion parameters and HER2positive tumors in this study, and other studies have reported the same results (3,4). Those studies included a relatively small number of malignancies with heterogeneous tumor profiles and used different analysis methods; therefore, a prospective study with a consistent design is required in a large homogeneous tumor group. DCE-MRI parameters obtained from different analysis approaches have shown different correlations with angiogenesis markers (21). As tumor vascularity or cellularity is heterogeneous, enhancement kinetics and ADC values are dependent in part on the placement of the ROI within the tumor. Therefore, ROI analysis is inevitably operator dependent, but histogram analysis covering the whole tumor is relatively operator independent. In the present study, histogram analysis showed greater association to prognostic factors than ROI analysis, reflecting overall tumor kinetics or tumor niche. This study has several limitations. First, it was retrospectively performed. Second, the number of cases was relatively small and we did multiple comparisons using many variables. Third, although only IDC lesions were included in this study, many of which had variable amounts of ductal carcinoma in situ (DCIS) component (48 cases of 52 cases) or tumor necrosis (three cases of 52 cases), which can be potential confounding factors. Fourth, comparison of direct histopathologic 10
Academic Radiology, Vol ■, No ■, ■■ 2016
angiogenesis markers, such as microvessel density (22), vascular endothelial growth factor, and hypoxia-inducible factor 1-alpha (23) to perfusion parameters or permeability parameters, was not performed. In conclusion, among MRI perfusion parameters, significant associations were observed between iAUC and tumor size and between ve median and PR positivity in IDC of the breast. Mean and median ADC values showed positive association with tumors with HER2 positivity and HER2-enriched subtype. In addition, histogram analysis of perfusion parameters and ADC values in the entire tumor volume showed greater heterogeneity of tumor kinetics and physiology than ROI analysis. These MR perfusion or ADC histogram parameters would be applicable in breast cancer patients who received neoadjuvant chemotherapy to predict treatment response, as other parts of body such as kidney or ovary (24,25).
ACKNOWLEDGMENTS This work was based on the Olea Sphere 2.3 (Olea Medical, France) and OncoTreat (Siemens Healthcare, works in progress) postprocessing software packages. The statistical consultation was supported by a grant of the Korean Health Technology R&D Project, Ministry of Health & Welfare, Republic of Korea (HI14C1731). This study was supported by the Research Fund of Seoul St. Mary’s Hospital, The Catholic University of Korea.
REFERENCES 1. Goldhirsch A, Glick JH, Gelber RD, et al. Meeting highlights: international expert consensus on the primary therapy of early breast cancer 2005. Ann Oncol 2005; 16:1569–1583. 2. Kim EJ, Kim SH, Park GE, et al. Histogram analysis of apparent diffusion coefficient at 3.0t: correlation with prognostic factors and subtypes of invasive ductal carcinoma. J Magn Reson Imaging 2015; doi:10.1002/jmri.24934. 3. Kim JY, Kim SH, Kim YJ, et al. Enhancement parameters on dynamic contrast enhanced breast MRI: do they correlate with prognostic factors and subtypes of breast cancers? Magn Reson Imaging 2015; 33:72– 80. 4. Koo HR, Cho N, Song IC, et al. Correlation of perfusion parameters on dynamic contrast-enhanced MRI with prognostic factors and subtypes of breast cancers. J Magn Reson Imaging 2012; 36:145–151. 5. Parker GM, Buckley D. Tracer kinetic modelling for T1-weighted DCEMRI. In: Jackson A, Buckley D, Parker GM, eds. Dynamic contrastenhanced magnetic resonance imaging in oncology. Berlin Heidelberg: Springer, 2005; 81–92. 6. Tofts PS, Brix G, Buckley DL, et al. Estimating kinetic parameters from dynamic contrast-enhanced T(1)-weighted MRI of a diffusable tracer: standardized quantities and symbols. J Magn Reson Imaging 1999; 10:223– 232. 7. Cheang MC, Chia SK, Voduc D, et al. Ki67 index, HER2 status, and prognosis of patients with luminal B breast cancer. J Natl Cancer Inst 2009; 101:736–750. 8. Goldhirsch A, Winer EP, Coates AS, et al. Personalizing the treatment of women with early breast cancer: highlights of the St Gallen International Expert Consensus on the Primary Therapy of Early Breast Cancer 2013. Ann Oncol 2013; 24:2206–2223. 9. Pickles MD, Manton DJ, Lowry M, et al. Prognostic value of pretreatment DCE-MRI parameters in predicting disease free and overall survival for breast cancer patients undergoing neoadjuvant chemotherapy. Eur J Radiol 2009; 71:498–505.
Academic Radiology, Vol ■, No ■, ■■ 2016
10. Walker-Samuel S, Leach MO, Collins DJ. Evaluation of response to treatment using DCE-MRI: the relationship between initial area under the gadolinium curve (IAUGC) and quantitative pharmacokinetic analysis. Phys Med Biol 2006; 51:3593–3602. 11. Horwitz KB, McGuire WL. Specific progesterone receptors in human breast cancer. Steroids 1975; 25:497–505. 12. Razek AA, Gaballa G, Denewer A, et al. Invasive ductal carcinoma: correlation of apparent diffusion coefficient value with pathological prognostic factors. NMR Biomed 2010; 23:619–623. 13. Park SH, Choi HY, Hahn SY. Correlations between apparent diffusion coefficient values of invasive ductal carcinoma and pathologic factors on diffusion-weighted MRI at 3.0 Tesla. J Magn Reson Imaging 2015; 41:175–182. 14. Kim SH, Cha ES, Kim HS, et al. Diffusion-weighted imaging of breast cancer: correlation of the apparent diffusion coefficient value with prognostic factors. J Magn Reson Imaging 2009; 30:615–620. 15. Ludovini V, Sidoni A, Pistola L, et al. Evaluation of the prognostic role of vascular endothelial growth factor and microvessel density in stages I and II breast cancer patients. Breast Cancer Res Treat 2003; 81:159– 168. 16. Black R, Prescott R, Bers K, et al. Tumour cellularity, oestrogen receptors and prognosis in breast cancer. Clin Oncol 1983; 9:311– 318. 17. Kumar R, Yarmand-Bagheri R. The role of HER2 in angiogenesis. Semin Oncol 2001; 28(5 suppl 16):27–32.
PERFUSION PARAMETERS IN BREAST CANCER
18. Martincich L, Deantoni V, Bertotto I, et al. Correlations between diffusionweighted imaging and breast cancer biomarkers. Eur Radiol 2012; 22:1519–1528. 19. Le Bihan D, Breton E, Lallemand D, et al. Separation of diffusion and perfusion in intravoxel incoherent motion MR imaging. Radiology 1988; 168:497–505. 20. Dvorak HF, Nagy JA, Dvorak JT, et al. Identification and characterization of the blood vessels of solid tumors that are leaky to circulating macromolecules. Am J Pathol 1988; 133:95–109. 21. Su MY, Cheung YC, Fruehauf JP, et al. Correlation of dynamic contrast enhancement MRI parameters with microvessel density and VEGF for assessment of angiogenesis in breast cancer. J Magn Reson Imaging 2003; 18:467–477. 22. Weidner N. Intratumor microvessel density as a prognostic factor in cancer. Am J Pathol 1995; 147:9–19. 23. Carmeliet P, Jain RK. Angiogenesis in cancer and other diseases. Nature 2000; 407:249–257. 24. Bharwani N, Miquel ME, Powles T, et al. Diffusion-weighted and multiphase contrast-enhanced MRI as surrogate markers of response to neoadjuvant sunitinib in metastatic renal cell carcinoma. Br J Cancer 2014; 110:616– 624. 25. Kyriazi S, Collins DJ, Messiou C, et al. Metastatic ovarian and primary peritoneal cancer: assessing chemotherapy response with diffusionweighted MR imaging—value of histogram analysis of apparent diffusion coefficients. Radiology 2011; 261:182–192.
11
