Association of mammography and ultrasound features with MammaPrint in patients with estrogen receptor-positive,HER2-negative,node-positive invasive breast cancer

Abstract

Background

MammaPrint is a 70-gene signature microarray assay that predicts the likelihood of recurrence of breast cancer and chemotherapeutic benefits.

Purpose

To investigate the association between mammography and ultrasound (US) features and MammaPrint results in patients with estrogen receptor (ER)-positive, HER2-negative, node-positive invasive breast cancer, and to identify the predictive factors for high risk of recurrence.

Material and Methods

This retrospective study included 251 patients with ER-positive, HER2-negative, 1–3 node-positive invasive breast cancer. Mammography and US findings were reviewed according to the BI-RADS criteria. The association between MammaPrint results and the clinicopathological and imaging features was evaluated. Logistic regression analysis was performed to identify independent predictors for high risk of recurrence.

Results

Of the patients, 143 (57.0%) and 108 (43.0%) had low and high risks for recurrence on MammaPrint, respectively. Young age (odds ratio [OR] 1.08; 95% confidence interval (CI) 1.04–1.12; P<0.001), posterior enhancement on US (OR 2.45; 95% CI 1.16–5.20; P = 0.019), absence of posterior shadowing on US (OR 3.19; 95% CI 1.17–8.62; P = 0.023), high histologic grade (OR 113.36; 95% CI 6.79–1893.53; P = 0.001), and high Ki-67 level (OR 4.90; 95% CI 2.62–9.17; P<0.001) were independently associated with high risk of recurrence on multivariate logistic regression analysis.

Conclusion

Posterior features in US may predict a high risk of recurrence in patients with ER-positive, HER2-negative, node-positive invasive breast cancer, which may be useful in enhancing the diagnostic value of MammaPrint and aid in the decision-making process regarding treatment.

Keywords

Breast cancer mammography MammaPrint ultrasound

Introduction

Breast cancer is a heterogeneous disease, and the treatment decisions differ according to the tumor characteristics. Women with early-stage breast cancer are often treated with adjuvant systemic therapy comprising endocrine therapy and chemotherapy. Accurate identification of the patient’s disease may prevent over-treatment with chemotherapy, and several gene expression profiling methods have been thus developed to aid in the decision-making process. For hormone receptor-positive, human epidermal growth factor 2 (HER2)-negative, early-stage breast cancer, Oncotype DX (1,2), MammaPrint (3 –6), PAM50 (7), Breast Cancer Index (8,9), and EndoPredict (10,11) have been studied for their value in predicting the prognosis and improving disease management.

The MammaPrint® (Agendia, Amsterdam, The Netherlands) is a 70-gene signature microarray assay that categorizes patients as having high or low risks of recurrence, based on the risk of distant metastasis at 5 and 10 years after treatment (4,5). The prospective randomized MINDACT (Microarray in Node-negative and 1 to 3 positive lymph node Disease may Avoid Chemotherapy) trial validated the efficacy of MammaPrint; in this trial, 46% of the patients at high risk of distant recurrence based on the clinical factors were classified as genomically low-risk patients based on the MammaPrint results and did not receive adjuvant chemotherapy (3). Furthermore, the IMPACt (Impact of MammaPrint on Adjuvant and Neoadjuvant Treatment in Breast Cancer Patients: A Prospective Registry) trial reported that 88.5% of the treatment plans were concordant with the MammaPrint results, along with a 60% decrease in the treatment recommendation for patients whose clinical features indicated high risk but were deemed as low risk by MammaPrint (6).

Furthermore, several studies have used radiological imaging features in combination with the results of the multigene expression assays such as the Oncotype DX (12 –16). Yepes et al. (15) demonstrated that the detection of a mass with pleomorphic micro-calcifications on mammography or the presence of posterior enhancement on ultrasound (US) may predict an intermediate-to-high recurrence score. A previous study identified parallel orientation and tumor roundness as independent variables for predicting a high risk of recurrence in patients undergoing Oncotype DX (12), and Dialani et al. (14) reported that a high recurrence score in Oncotype DX was correlated with the presence of a well-circumscribed oval mass on mammography, posterior enhancement and vascularity on US, and a lobulated mass on magnetic resonance imaging (MRI). Another study investigated the relationship between computer-extracted breast MRI phenotypes and multigene assays such as the MammaPrint, Oncotype DX, and PAM50, and reported that tumors with a high risk of recurrence tended to be large and showed more heterogeneous enhancement patterns (13). However, to the best of our knowledge, previous studies have not investigated the correlation between mammography and US characteristics of breast cancer patients and their MammaPrint results. We thus aimed to investigate the association between the mammography and US features and MammaPrint results in patients with ER-positive, HER2-negative, node-positive invasive breast cancer and to identify the predictive factors for high risk of recurrence.

Material and Methods

The institutional review board of Asan Medical Center approved this study, and waived the requirement for informed consent based on the retrospective nature of this study.

Study population

The study population included 271 patients with ER-positive, HER2-negative breast cancer with 1–3 positive lymph nodes who underwent MammaPrint at our institution between January 2018 and December 2019. Of the patients, the following patients were excluded: patients who underwent surgery elsewhere (n = 6); patients who underwent excisional biopsy at another hospital (n = 5); patients who received neoadjuvant chemotherapy (n = 3); patients with a history of breast cancer (n = 3); male patients (n = 2); and patients who did not have US images (n = 1). As a result, the study population comprised 251 women.

Mammography and US image acquisition

Mammography was performed using the full-field digital mammography system (Senographe Essential, GE Healthcare, Milwaukee, WI, USA; Selenia Dimensions, Hologic, Bedford, MA, USA; Mammomat Inspiration, Siemens Healthcare, Erlangen, Germany). Bilateral craniocaudal and mediolateral oblique views were routinely obtained. B-mode US was performed using high-resolution US equipment with a 5–12-MHz or 22–2-MHz linear array transducer (IU22 or EPIQ: Philips Medical Systems, Bothell, WA, USA) by 11 board-certified radiologists with 5–24 years of experience in breast imaging.

Image analysis

Two radiologists (H.S. and W.J.C., with 5 and 10 years of experience in breast imaging, respectively) independently reviewed the preoperative mammographic and US images according to the guidelines of the American College of Radiology Breast Imaging Reporting and Data System (BI-RADS), 5th edition (3). To reduce inter-observer variability, another radiologist joined the interpretation in case of a discrepancy between the two readers, and all three radiologists reached a consensus in describing the imaging findings. The radiologists were blinded to the clinicopathological findings. The mammographic breast densities were classified as either fatty (almost entirely fatty or with scattered areas of fibroglandular density) or dense (heterogeneously or extremely dense). Mammographic features (e.g. lesion type, mass shape, margin, density, and morphology and distribution of calcification) and US features (e.g. mass shape, orientation, margin, echo pattern, and posterior features) were also evaluated.

MammaPrint and pathological data analysis

Samples from surgically excised tumors were analyzed using the MammaPrint microarray assay (Agendia, Amsterdam, The Netherlands). We reviewed the MammaPrint results, which provide index scores in the range of –1.000 and 1.000, and categorized the tumors as having low (index score = 0.001–1.000) or high (index score = –1.000 to 0.000) risk of recurrence. By reviewing the postoperative pathological reports, we also assessed the pathological data including the histological tumor type, size, histologic grade, lymph node metastasis, lymphovascular invasion, expression of epidermal growth factor receptor (EGFR), cytokeratin (CK) 5/6, p53, and Ki-67 proliferation.

Statistical analysis

Statistical analyses were performed using the IBM SPSS Statistics for Windows, version 21 (IBM Corp., Armonk, NY, USA). P < 0.05 was considered statistically significant. The association between MammaPrint results and clinicopathological and imaging features was compared using Student’s t-test or Kruskal–Wallis test for continuous variables and chi-square test or Fisher’s exact test for categorical variables. Logistic regression analyses were performed to identify clinicopathological and imaging features that are independently associated with a high risk of recurrence with respect to the MammaPrint results. Variables with P < 0.05 in univariate analysis were included in the multivariate logistic regression analysis, with backward elimination. Receiver operating characteristic (ROC) analysis was performed to assess the diagnostic accuracy of the logistic model.

Results

Association of clinicopathological data and MammaPrint results

Of the 251 patients, 143 (57.0%) and 108 (43.0%) were at low and high risk of recurrence, respectively, based on the MammaPrint results. Table 1 summarizes the clinicopathological characteristics of the patients. The patients were aged 31–79 years (mean age = 51 years) and those in the high-risk group were significantly younger (P < 0.001). The mean tumor size was 2.3 cm (range = 0.4–6.5 cm) and the majority of the tumors were classified as invasive ductal carcinomas (n = 223, 88.8%), while the other histologic types included invasive lobular carcinoma (n = 19), mucinous carcinoma (n = 6), invasive micropapillary carcinoma (n = 2), and tubular carcinoma (n = 1). All patients presented with node-positive breast cancer with 1–3 positive lymph node metastases. In terms of histologic grade, 15 (6.0%), 220 (87.6%), and 16 (6.4%) tumors were of low, intermediate, and high histologic grade, respectively. The high-risk group was significantly associated with high histologic grade (P < 0.001), lymphovascular invasion (P = 0.001), and high levels of Ki-67 (P < 0.001) compared with the low-risk group.

Table 1.

Patient and tumor characteristics according to the risk of recurrence.

Variable	All	Low risk	High risk	P
Number of patients	251 (100)	143 (57.0)	108 (43.0)
Patient age (years)	51.0 ± 9.3	53.4 ± 9.1	47.8 ± 8.5	<0.001
Surgery type				0.442
Breast conservation	194 (77.3)	108 (75.5)	86 (79.6)
Mastectomy	57 (22.7)	35 (24.5)	22 (20.4)
Tumor size (cm)	2.3 ± 1.0	2.2 ± 1.0	2.4 ± 0.9	0.029
Histologic type				0.053
Ductal	223 (88.8)	125 (87.4)	98 (90.7)
Lobular	19 (7.6)	15 (10.5)	4 (3.7)
Others*	9 (3.6)	3 (2.1)	6 (5.6)
Histologic grade				<0.001
Low	15 (6.0)	13 (9.1)	2 (1.9)
Intermediate	220 (87.6)	129 (90.2)	91 (84.3)
High	16 (6.4)	1 (0.7)	15 (13.9)
Lymphovascular invasion				0.001
Absent	129 (51.4)	86 (60.1)	43 (39.8)
Present	122 (48.6)	57 (39.9)	65 (60.2)
EGFR				0.579
Negative	248 (98.8)	142 (99.3)	106 (98.2)
Positive	3 (1.2)	1 (0.7)	2 (1.9)
CK5/6				0.430
Negative	250 (99.6)	143 (100)	107 (99.1)
Positive	1 (0.4)	0 (0)	1 (0.9)
p53				0.071
Negative	95 (37.8)	61 (42.7)	34 (31.5)
Positive	156 (62.2)	82 (57.3)	74 (68.5)
Ki-67				<0.001
Low (<14%)	142 (56.6)	107 (74.8)	35 (32.4)
High (≥14%)	109 (43.4)	36 (25.2)	73 (67.6)

Values are given as n (%) or mean ± SD.

*Includes mucinous carcinoma, invasive micropapillary carcinoma, and tubular carcinoma.

CK, cytokeratin; EGFR, epidermal growth factor receptor; SD, standard deviation.

Association of imaging features and MammaPrint results

The mammographic and US findings based on the BI-RADS criteria are summarized in Tables 2 and 3. Twenty-three tumors (9.2%) were not detected on mammography, while 153 (61.0%) tumors appeared as masses, 15 (6.0%) tumors presented with only calcification, and 60 (23.9%) tumors appeared as masses with calcification. The high-risk group was significantly associated with dense breast tissue (P = 0.027) and indistinct mass margin (P = 0.021). No significant differences were observed between the low- and high-risk groups according to mass shape, density, and morphology and distribution of calcification on mammography. On US, all lesions appeared as masses, and the posterior features were significantly different according to the risk of recurrence (P < 0.001), whereas the mass shape, orientation, margin, and echo pattern in US did not significantly differ between the two groups. Representative images of patients in the low- and high-risk groups are shown in Figs. 1 and 2, respectively.

Table 2.

Mammography features according to the risk of recurrence.

Variable	All	Low risk	High risk	P
Parenchymal density				0.027
Fatty	32 (12.7)	24 (16.8)	8 (7.4)
Dense	219 (87.3)	119 (83.2)	100 (92.6)
Lesion type				0.401
No visualization	23 (9.2)	15 (10.5)	8 (7.4)
Mass	153 (61.0)	90 (62.9)	63 (58.3)
Calcification only	15 (6.0)	6 (4.2)	9 (8.3)
Mass with calcification	60 (23.9)	32 (22.4)	28 (25.9)
Mass shape				0.167
Oval	12 (5.6)	6 (4.9)	6 (6.6)
Round	10 (4.7)	3 (2.5)	7 (7.7)
Irregular	191 (89.7)	113 (92.6)	78 (85.7)
Mass margin				0.021
Indistinct	75 (35.2)	35 (28.7)	40 (44.0)
Spiculated	138 (64.8)	87 (71.3)	51 (56.0)
Mass density				0.166
High	103 (48.4)	54 (44.3)	49 (53.9)
Equal	110 (51.6)	68 (55.7)	42 (46.2)
Calcification morphology				0.311
Amorphous	30 (40.0)	19 (50.0)	11 (29.7)
Coarse heterogeneous	12 (16.0)	6 (15.8)	6 (16.2)
Fine pleomorphic	30 (40.0)	12 (31.6)	18 (48.7)
Fine linear branching	3 (4.0)	1 (2.6)	2 (5.4)
Calcification distribution				0.128
Diffuse	3 (4.0)	2 (5.3)	1 (2.7)
Regional	11 (14.7)	9 (23.7)	2 (5.4)
Grouped	48 (64.0)	22 (57.9)	26 (70.3)
Linear	1 (1.3)	0 (0)	1 (2.7)
Segmental	12 (16.0)	5 (13.2)	7 (18.9)

Values are given as n (%).

Table 3.

Ultrasound features according to the risk of recurrence.

Variable	All	Low risk	High risk	P
Mass shape				0.470
Oval	16 (6.4)	11 (7.7)	5 (4.6)
Round	7 (2.8)	5 (3.5)	2 (1.9)
Irregular	228 (90.8)	127 (88.8)	101 (93.5)
Mass orientation				0.946
Parallel	205 (81.7)	117 (81.8)	88 (81.5)
Non-parallel	46 (18.3)	26 (18.2)	20 (18.5)
Mass margin				0.589
Circumscribed	2 (0.8)	1 (0.7)	1 (0.9)
Indistinct	102 (40.6)	52 (36.4)	50 (46.3)
Angular	27 (10.8)	17 (11.9)	10 (9.3)
Microlobulated	16 (6.4)	9 (6.3)	7 (6.5)
Spiculated	104 (41.4)	64 (44.8)	40 (37.0)
Mass echo pattern				0.465
Hypoechoic	220 (87.6)	122 (85.3)	98 (90.7)
Isoechoic	4 (1.6)	3 (2.1)	1 (0.9)
Heterogeneous	27 (10.8)	18 (12.6)	9 (8.3)
Mass posterior feature				< 0.001
None	101 (40.2)	59 (41.3)	42 (38.9)
Enhancement	64 (25.5)	22 (15.4)	42 (38.9)
Shadowing	45 (17.9)	37 (25.9)	8 (7.4)
Combined	41 (16.3)	25 (17.5)	16 (14.8)

Values are given as n (%).

Fig. 1.

A 64-year-old woman with a MammaPrint result of 0.2 (low risk). (a) Mammography image showing spiculated irregular-shaped mass without calcification (arrow) in the scattered areas of fibroglandular density of the right breast. (b) Breast ultrasound image showing spiculated mass with no posterior feature (arrow).

Fig. 2.

A 44-year-old woman with a MammaPrint result of –0.2 (high risk). (a) Mammography image showing indistinct irregular-shaped mass without calcification (arrow) in the extremely dense parenchymal tissues of the right breast. (b) Breast ultrasound image showing irregular mass with posterior enhancement (arrow).

Table 4.

Clinicopathological and imaging features associated with high risk of recurrence based on logistic regression analysis.

	Univariate analysis		Multivariate analysis
Characteristics	Unadjusted OR (95% CI)	P	Adjusted OR (95% CI)	P
Age	0.93 (0.90–0.96)	<0.001	0.93 (0.90–0.97)	<0.001
MG parenchymal density		0.032
Fatty	Reference
Dense	2.52 (1.09–5.86)
US posterior feature		<0.001		0.001
None	Reference		Reference
Enhancement	2.69 (1.40–5.14)	0.003	2.45 (1.16–5.20)	0.019
Shadowing	0.30 (0.13–0.72)	0.007	0.31 (0.12–0.85)	0.023
Combined	0.90 (0.43–1.89)	0.779	0.72 (0.29–1.79)	0.483
Histologic grade		0.002		0.004
Low	Reference		Reference
Intermediate	4.59 (1.01–20.81)	0.048	5.49 (1.04–29.01)	0.045
High	97.5 (7.90–1203.00)	<0.001	113.36 (6.79–1893.53)	0.001
Lymphovascular invasion		0.002
Absent	Reference
Present	2.28 (1.37–3.80)
Ki-67		<0.001		<0.001
Low	Reference		Reference
High	6.20 (3.57–10.78)		4.90 (2.62–9.17)

CI, confidence interval; MG, mammography; OR, odds ratio; US, ultrasound.

Logistic regression analysis

Univariate analyses showed that the high-risk group according to MammaPrint results was significantly associated with age (P < 0.001), mammographic parenchymal density (P = 0.032), posterior features in US (P < 0.001), histologic grade (P = 0.002), lymphovascular invasion (P = 0.002), and level of Ki-67 (P < 0.001). Multivariate logistic regression analysis showed that younger age (odds ratio (OR) = 1.08; 95% confidence interval (CI) = 1.04–1.12; P < 0.001), posterior enhancement on US (OR = 2.45; 95% CI = 1.16–5.20; P = 0.019), absence of posterior shadowing on US (OR = 3.19; 95% CI = 1.17–8.62; p = 0.023), high histologic grade (OR = 113.36; 95% CI = 6.79–1893.53; P = 0.001), and high levels of Ki-67 (OR = 4.90; 95% CI = 2.62–9.17; P < 0.001) were independent factors associated with high recurrence score. The A_z values for age, posterior feature, histologic grade, and levels of Ki-67 were 0.67, 0.54, 0.60, and 0.71, respectively (Fig. 3). Also, the A_z increased to 0.84 in the model in which all independent variables were used.

Fig. 3.

Receiver operating characteristic curve determined by multivariate logistic regression analysis for distinguishing high recurrence scores from low or intermediate recurrence scores. Data in parentheses indicate the A_z values for each variable.

Discussion

Here, we observed that patients with ER-positive, HER2-negative, node-positive invasive breast cancer have different clinicopathological and imaging features according to the risk of recurrence as determined by the MammaPrint results. Posterior enhancement and absence of posterior shadowing in US images were independently associated with a high risk of recurrence. Additionally, young age, high histologic grade, and high levels of Ki-67 were more frequently observed in the high-risk group.

The posterior features in US images represent the attenuation characteristics of the mass. In our study, posterior enhancement (OR = 2.45; P = 0.019) and absence of posterior shadowing (OR = 3.19; P = 0.023) were identified as independent factors for the high-risk group. Posterior acoustic enhancement was reported to be associated with a high histological grade (17), and this finding corroborated the findings of previous studies in which posterior enhancement was correlated with the Oncotype DX results in the high-risk group (12,14,15). As high-grade tumors have higher mitotic rates and higher cellularity, they show less attenuation in the US waves, which leads to posterior enhancement. In contrast, low-grade tumors may show posterior shadowing as they are slow-growing and have lower mitotic rates, thus leading to marked differences in the acoustic impedance of tumor interfaces (18).

On mammography, the high-risk group showed significantly higher breast density (P = 0.027) and a higher number of masses with indistinct margins (P = 0.021), although these were not identified as independent predictive factors for high risk of recurrence in logistic regression analysis. Masses with indistinct margins are associated with a higher histologic grade and presence of extensive intraductal component (17). Woodard et al. (19) observed that while the indistinct mass margin had the highest Oncotype DX recurrence score, the breast density was inversely correlated with the Oncotype DX recurrence score. In our study, spiculated margins were associated with the low-risk group, which is consistent with the findings of a previous study in which spiculated margins were associated with low-grade breast tumors (20) and a more favorable prognosis (21). Increased breast density is a known risk factor for breast cancer (22), and although there is controversy regarding the association between breast density and breast cancer-specific survival, some studies have reported that dense breasts were associated with decreased breast cancer-specific survival (23,24).

Analysis of the relationships among imaging features, gene expression profiles, and patient outcomes may provide important information for individual patients. Moreover, the Oncotype DX assay has been widely used to predict the prognosis of patients with ER-positive, HER2-negative, early-stage breast cancer. However, while the recurrence score classifies tumors into low- (<18), intermediate- (18 –30), or high-risk (≥31) groups, there is controversy regarding the intermediate-risk group with respect to the efficacy of adjuvant chemotherapy. Sparano et al. (25) reported similar efficacy of the adjuvant endocrine therapy and chemo-endocrine therapy in patients with mid-range Oncotype DX recurrence score, although limited benefits of chemotherapy were observed in some women aged ≤50 years.

The MammaPrint assay, as opposed to the Oncotype DX, classifies patients into low- or high-risk groups, and may aid clinicians in decision-making regarding additional treatment. The MammaPrint assay is a prognostic parameter in the St. Gallen and European Society of Medical Oncology (ESMO) guidelines (26,27). Mook et al. (28) demonstrated that the MammaPrint assay may be used to identify low-risk patients among those with 1–3 node-positive breast cancer, who are traditionally considered high-risk patients for recurrence based on the conventional histopathological evaluation; specifically, their study showed that the 70-gene signature was the most powerful predictor of breast cancer-specific survival even after adjustment for clinicopathological factors.

In the present study, clinicopathological variables such as age, histologic grade, and levels of Ki-67 were associated with the high-risk group. The high histologic grade variable had the highest OR for predicting the high-risk group in multivariate analysis. These results are consistent with those of previous studies in which the histologic grade was reported as a strong prognostic and predictive factor for the outcomes of patients with breast cancer (29,30). Ki-67 is a proliferation marker that serves as an independent prognostic factor (31) and was included in the Oncotype DX assay (2). Moreover, younger women with breast cancer may show aggressive tumors with worse prognoses (32).

The present study has several limitations. First, this was a retrospective study involving a limited number of patients from a single tertiary referral center. Second, US imaging analysis was based on static images and real-time evaluation was not possible. Finally, we only evaluated the relationship between imaging features and the MammaPrint result, and the actual tumor recurrence in patients was not evaluated. In a future study, correlating the imaging features with tumor recurrence and developing a model for predicting high recurrence scores may further aid in the decision for treatment.

In conclusion, we report that young age, posterior enhancement on US imaging, absence of posterior shadowing on US imaging, high histologic grade, and high levels of Ki-67 were independent predictors for high recurrence score in patients with ER-positive, HER2-negative, node-positive invasive breast cancer. The posterior features in US images may predict a high risk of recurrence in such patients, thus providing added diagnostic value and aiding in decision-making regarding additional treatment.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Woo Jung Choi

Joo Hee Cha

Eun Young Chae

References

Paik

Tang

Shak

, et al. Gene expression and benefit of chemotherapy in women with node-negative, estrogen receptor-positive breast cancer. J Clin Oncol 2006; 24:3726–3734.

Paik

Shak

Tang

, et al. A multigene assay to predict recurrence of tamoxifen-treated, node-negative breast cancer. N Engl J Med 2004; 351:2817–2826.

Cardoso

van’t Veer

Bogaerts

, et al. 70-gene signature as an aid to treatment decisions in early-stage breast cancer. N Engl J Med 2016; 375:717–729.

van’t Veer

Dai

van de Vijver

, et al. Gene expression profiling predicts clinical outcome of breast cancer. Nature 2002; 415:530–536.

van de Vijver

van’t Veer

, et al. A gene-expression signature as a predictor of survival in breast cancer. N Engl J Med 2002; 347:1999–2009.

Soliman

Shah

Srkalovic

, et al. MammaPrint guides treatment decisions in breast cancer: results of the IMPACt trial. BMC Cancer 2020; 20:81.

Filipits

Nielsen

Rudas

, et al. The PAM50 risk-of-recurrence score predicts risk for late distant recurrence after endocrine therapy in postmenopausal women with endocrine-responsive early breast cancer. Clin Cancer Res 2014; 20:1298–1305.

Dubsky

Filipits

Jakesz

, et al. EndoPredict improves the prognostic classification derived from common clinical guidelines in ER-positive, HER2-negative early breast cancer. Ann Oncol 2013; 24:640–647.

Filipits

Rudas

Jakesz

, et al. A new molecular predictor of distant recurrence in ER-positive, HER2-negative breast cancer adds independent information to conventional clinical risk factors. Clin Cancer Res 2011; 17:6012–6020.

10.

Zhang

Schnabel

Schroeder

, et al. Breast cancer index identifies early-stage estrogen receptor-positive breast cancer patients at risk for early- and late-distant recurrence. Clin Cancer Res 2013; 19:4196–4205.

11.

Sgroi

Sestak

Cuzick

, et al. Prediction of late distant recurrence in patients with oestrogen-receptor-positive breast cancer: a prospective comparison of the breast-cancer index (BCI) assay, 21-gene recurrence score, and IHC4 in the TransATAC study population. Lancet Oncol 2013; 14:1067–1076.

12.

Chae EY, Moon WK, Kim HH, et al. Association between ultrasound features and the 21-gene recurrence score assays in patients with oestrogen receptor-positive, HER2-negative, invasive breast cancer. PLoS One 2016;11:e0158461.

13.

Zhu

Burnside

, et al. MR imaging radiomics signatures for predicting the risk of breast cancer recurrence as given by research versions of MammaPrint, Oncotype DX, and PAM50 gene assays. Radiology 2016; 281:382–391.

14.

Dialani

Gaur

Mehta

, et al. Prediction of low versus high recurrence scores in estrogen receptor-positive, lymph node-negative invasive breast cancer on the basis of radiologic-pathologic features: comparison with Oncotype DX test recurrence scores. Radiology 2016; 280:370–378.

15.

Yepes

Romilly

Collado-Mesa

, et al. Can mammographic and sonographic imaging features predict the Oncotype DX recurrence score in T1 and T2, hormone receptor positive, HER2 negative and axillary lymph node negative breast cancers? Breast Cancer Res Treat 2014; 148:117–123.

16.

Kim

Hwangbo

, et al. Diffusion-weighted MRI of estrogen receptor-positive, HER2-negative, node-negative breast cancer: association between intratumoral heterogeneity and recurrence risk. Eur Radiol 2020; 30:66–76.

17.

Shin HJ, Kim HH, Huh MO, et al. Correlation between mammographic and sonographic findings and prognostic factors in patients with node-negative invasive breast cancer. Br J Radiol 2011;84:19–30.

18.

Irshad

Leddy

Pisano

, et al. Assessing the role of ultrasound in predicting the biological behavior of breast cancer. AJR Am J Roentgenol 2013; 200:284–290.

19.

Woodard

Ray

Joe

, et al. Qualitative radiogenomics: association between Oncotype DX test recurrence score and BI-RADS mammographic and breast MR imaging features. Radiology 2018; 286:60–70.

20.

Stavros

Thickman

Rapp

, et al. Solid breast nodules: use of sonography to distinguish between benign and malignant lesions. Radiology 1995; 196:123–134.

21.

Evans

Pinder

James

, et al. Is mammographic spiculation an independent, good prognostic factor in screening-detected invasive breast cancer? AJR Am J Roentgenol 2006; 187:1377–1380.

22.

Price

Hargreaves

Lipson

, et al. The California breast density information group: a collaborative response to the issues of breast density, breast cancer risk, and breast density notification legislation. Radiology 2013; 269:887–892.

23.

Olsson

Sartor

Borgquist

, et al. Breast density and mode of detection in relation to breast cancer specific survival: a cohort study. BMC Cancer 2014; 14:229.

24.

Chiu

Duffy

Yen

, et al. Effect of baseline breast density on breast cancer incidence, stage, mortality, and screening parameters: 25-year follow-up of a Swedish mammographic screening. Cancer Epidemiol Biomarkers Prev 2010; 19:1219–1228.

25.

Sparano

Gray

Makower

, et al. Adjuvant chemotherapy guided by a 21-gene expression assay in breast cancer. N Engl J Med 2018; 379:111–121.

26.

Senkus

Kyriakides

Ohno

, et al. Primary breast cancer: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol 2015; 26:v8–30.

27.

Coates

Winer

Goldhirsch

, et al. Tailoring therapies-improving the management of early breast cancer: St Gallen International Expert Consensus on the Primary Therapy of Early Breast Cancer 2015. Ann Oncol 2015; 26:1533–1546.

28.

Mook

Schmidt

Viale

, et al. The 70-gene prognosis-signature predicts disease outcome in breast cancer patients with 1-3 positive lymph nodes in an independent validation study. Breast Cancer Res Treat 2009; 116:295–302.

29.

Rakha

El-Sayed

Lee

, et al. Prognostic significance of Nottingham histologic grade in invasive breast carcinoma. J Clin Oncol 2008; 26:3153–3158.

30.

Desmedt

Haibe-Kains

Wirapati

, et al. Biological processes associated with breast cancer clinical outcome depend on the molecular subtypes. Clin Cancer Res 2008; 14:5158–5165.

31.

de Azambuja

Cardoso

de Castro

Jr. , et al. Ki-67 as prognostic marker in early breast cancer: a meta-analysis of published studies involving 12,155 patients. Br J Cancer 2007; 96:1504–1513.

32.

Nixon

Neuberg

Hayes

, et al. Relationship of patient age to pathologic features of the tumor and prognosis for patients with stage I or II breast cancer. J Clin Oncol 1994; 12:888–894.