Advertisement
Reviews

Imaging Features of HER2 Overexpression in Breast Cancer: A Systematic Review and Meta-analysis

Sjoerd G. Elias, Arthur Adams, Dorota J. Wisner, Laura J. Esserman, Laura J. van't Veer, Willem P.Th.M. Mali, Kenneth G.A. Gilhuijs and Nola M. Hylton
DOI: 10.1158/1055-9965.EPI-13-1170 Published August 2014

Abstract

Breast cancer imaging phenotype is diverse and may relate to molecular alterations driving cancer behavior. We systematically reviewed and meta-analyzed relations between breast cancer imaging features and human epidermal growth factor receptor type 2 (HER2) overexpression as a marker of breast cancer aggressiveness. MEDLINE and EMBASE were searched for mammography, breast ultrasound, magnetic resonance imaging (MRI), and/or [18F]fluorodeoxyglucose positron emission tomography studies through February 2013. Of 68 imaging features that could be pooled (85 articles, 23,255 cancers; random-effects meta-analysis), 11 significantly related to HER2 overexpression. Results based on five or more studies and robustness in subgroup analyses were as follows: the presence of microcalcifications on mammography [pooled odds ratio (pOR), 3.14; 95% confidence interval (CI), 2.46–4.00] or ultrasound (mass-associated pOR, 2.95; 95% CI, 2.34–3.71), branching or fine linear microcalcifications (pOR, 2.11; 95% CI, 1.07–4.14) or extremely dense breasts on mammography (pOR, 1.37; 95% CI, 1.07–1.76), and washout (pOR, 1.57; 95% CI, 1.11–2.21) or fast initial kinetics (pOR, 2.60; 95% CI, 1.43–4.73) on MRI all increased the chance of HER2 overexpression. Maximum [18F]fluorodeoxyglucose standardized uptake value (SUVmax) was higher upon HER2 overexpression (pooled mean difference, +0.76; 95% CI, 0.10–1.42). These results show that several imaging features relate to HER2 overexpression, lending credibility to the hypothesis that imaging phenotype reflects cancer behavior. This implies prognostic relevance, which is especially relevant as imaging is readily available during diagnostic work-up. Cancer Epidemiol Biomarkers Prev; 23(8); 1464–83. ©2014 AACR.

Introduction

Breast cancer is the most common type of cancer and the leading cause of cancer-related death in women worldwide (1). It is a heterogeneous disease, which can be appreciated by its diverse imaging appearance (2), its histologic and molecular classifications (3–5), and its correspondingly diverse disease course. One of the most clinically relevant molecular aberrations in breast cancer is overexpression of the human epidermal growth factor receptor type 2 (HER2). HER2 overexpression occurs in 15% to 25% of invasive breast cancers, and is associated with an intrinsic worse prognosis but good response to HER2-targeted therapies (6).

Some of the many effects on the cellular level of HER2 overexpression are increased cell proliferation, cell survival, mobility, and invasiveness, as well as neo-angiogenesis by increasing vascular endothelial growth factor production (7). These cellular processes and their clinical course provide evidence that HER2-overexpressing breast cancers behave distinctly from other breast cancers, which might drive macroscopic appearance and physiologic parameters. These phenomena may be potentially visible by clinical imaging modalities. As of now, the literature relating imaging features to HER2 overexpression in breast cancer is diverse, scattered over different scientific and clinical fields, and often based on small studies. The purpose of this study was to comprehensively review that literature and use meta-analysis techniques to formally quantify the relation between imaging features and HER2-positive breast cancer. We specifically focused on established clinical imaging modalities [mammography, breast ultrasound, magnetic resonance imaging (MRI), and [18F]fluorodeoxyglucose positron emission tomography (18F-FDG PET)].

Identification of imaging features related to HER2 overexpression in breast cancer could not only increase our biologic understanding, but may also have potential future clinical relevance. Existence of imaging features related to HER2 overexpression would for instance suggest prognostic value of breast cancer imaging phenotype. Furthermore, such features may have relevance in identifying potential sampling error in cases in which HER2 status is based on tumor biopsies, as practiced for neoadjuvant treatment indication.

Materials and Methods

Literature search and study selection

We performed a comprehensive systematic literature search of MEDLINE and EMBASE on February 8, 2013 using synonyms for HER2 and the imaging modalities of interest in combination with breast cancer (Fig. 1). The search was without restrictions. After combining the searches and duplicate removal, two researchers (S.G. Elias and A. Adams) independently performed all selection and data-extraction steps. First, we assessed titles and abstracts, excluding only articles deemed ineligible for full-text evaluation by both researchers. Then, we reviewed the full text of those remaining and subsequently excluded articles only upon consensus. We documented exclusion criteria as follows: (i) non-original data (e.g., reviews, editorials, and guidelines), (ii) preclinical studies (e.g., animal or in vitro studies), (iii) case reports (i.e., studies including less than 10 patients), (iv) non-primary breast cancer (e.g., imaging of breast cancer metastasis or lymph nodes), (v) experimental breast imaging modalities (e.g., optical mammography or breast elastography), (vi) treatment evaluation studies, and (vii) no imaging features described or evaluated. We then reviewed each selected article's references to identify any articles missed by the original search. Finally, articles with insufficient data for review (e.g., presenting only P values), or studies that presented identical data on the same patients were excluded (keeping the largest series). If unsure about duplicate data, we contacted the authors. As magnetic resonance spectroscopy (MRS) and diffusion-weighted imaging (DWI) are not established clinical imaging modalities but could be implemented rather easily given the clinical availability of MRI scanners, we did not exclude these modalities but show the results in the Supplementary Results.

Figure 1.
Figure 1.

Overview of the MEDLINE and EMBASE literature search and study selection process. aBoth searches performed on February 8, 2013; MEDLINE query: (her-2* OR her2* OR *erbb*) AND (mri OR mammograph* OR ultrasound OR (positron emission tomography) OR imaging) AND breast; EMBASE query: “epidermal growth factor receptor 2”/exp AND (“nuclear magnetic resonance imaging”/exp OR “mammography”/exp OR “echography”/exp OR “positron emission tomography”/exp) AND “breast cancer”/exp. Explanation of exclusion reasons, be.g., reviews, editorials, guidelines, and perspectives; ce.g., optical imaging and scintimammography; dfive studies reported both on mammography and ultrasonography, and three both on DCE-MRI and 18F-FDG PET.

Data extraction

First, we extracted study characteristics [e.g., study size, overall patient and tumor characteristics, HER2 assessment methodology, imaging acquisition details, use of the Breast Imaging-Reporting and Data System (BI-RADS), which improves comparability (2), and, more specifically related to the risk of bias, whether or not image assessment was blinded for HER2 status and whether the selection of patients may have introduced selection bias]. Then, we documented the numerical results of each imaging feature in relation to HER2 status [for categorical data by cross-tabulating absolute numbers, for continuous data using means and standard deviations (SD)]. If necessary, we matched different descriptors (e.g., for mammography, we combined studies describing “well-defined mass margins” and “smooth mass margins” to the overall descriptor “circumscribed margins”; Supplementary Tables S1–S3), combined groups [e.g., if features were compared between HER2-positive, triple-negative, and estrogen receptor (ER)–positive breast cancer, we combined the latter two groups], or approximated study data (e.g., using percentages and totals to derive absolute numbers). For group combination of categorical data, we used summation following cross-tabulation. For group combination of continuous data, we used inverse variance weighted pooling. Approximation of means and SD was necessary for several DWI and 18F-FDG PET studies that did not report these parameters using methods as outlined elsewhere (8, 9). All data-extraction steps were double-checked by two researchers (S.G. Elias and A. Adams).

Meta-analysis

For pooling of study results, we used DerSimonian–Laird random effects models to allow for between-study heterogeneity. For imaging features reported as categorical data, we estimated pooled odds ratios (pOR) to describe the relation between that feature and the chance of HER2 overexpression. All breast cancers not belonging to the imaging feature category of interest served as reference category in these analyses. For imaging features reported as continuous data, we estimated pooled mean differences between HER2-positive and -negative breast cancer. We assessed between-study heterogeneity by I2 statistics in combination with Cochran Q test for heterogeneity [denoted as P(Q)], and evaluated the impact of publication bias by inspecting funnel plot asymmetry in combination with Egger tests. We used forest plots to inspect individual study data and meta-analysis results.

Besides pooling of data using all available studies, we additionally performed meta-analyses in several predefined study subgroups to evaluate possible sources of between-study heterogeneity in the results: (i) excluding studies with data on pure ductal carcinoma in situ (DCIS), to focus on invasive cancer results only; (ii) excluding studies selecting participants based on receptor expression (e.g., studies that excluded ER-positive breast cancer, thereby contrasting triple-negative to HER2 positive breast cancer) as this may threaten generalizability of the results and/or might induce selection bias; (iii) focusing on studies with a specific imaging acquisition technique [analog vs. digital for mammography; studies including 3T vs. only 1.5T field strength for MRI, as well as higher spatial resolution MRI studies (i.e., sub-mm in plane resolution and smaller than 2-mm slice thickness)]; (iv) focusing on studies based on BI-RADS; and (v) excluding studies for which we had to approximate means and SDs.

We used R version 2.15.3 (R Foundation for Statistical Computing, Vienna, Austria) for all analyses including the packages rmeta and meta (10–12). We report pooled estimates in combination with 95% confidence intervals (CI) and used two-sided P values of <0.05 for statistical significance. The report and conduct of this meta-analysis satisfies the PRISMA Statement (13).

Results

Literature search and study selection

Figure 1 shows an overview of the literature search and study selection process. The search yielded 1,673 unique articles, of which we excluded 1,469 based on title and abstract (83% because of non-original data, preclinical, or case reports), and 117 following full-text screening, of which 47 (40%) were (neo)adjuvant treatment studies, without correlation between imaging features and HER2 status. Reference cross-checking of the 87 eligible articles yielded seven additional articles not initially identified (2 nonindexed breast ultrasound articles, refs. 14, 15; 1 MRS, ref. 16; and 4 18F-FDG PET articles, refs. 17–20). We subsequently excluded five articles because of insufficient data (20–24), and four because of patient overlap (19, 25–27). Thus, we selected 85 articles for our review (14–18, 28–107), representing 81 unique patient populations, totaling 23,159 patients with 23,255 breast cancers including 4,213 HER2-positives. Four populations gave rise to two separate publications on different imaging features and/or modalities (16, 54, 61, 66, 70, 75, 81, 104). A total of 33 articles reported on mammography (28–60), 12 on breast ultrasound (14, 15, 28, 33, 41, 52, 56, 61–65), 28 on MRI (16, 66–92), and 20 on 18F-FDG PET (17, 18, 70, 87, 91, 93–107). Of these, eight articles reported on two imaging modalities: five on both breast ultrasound and mammography (28, 33, 41, 52, 56), and three on both 18F-FDG PET and DWI (70, 87), or MRS (91). Table 1 shows the characteristics of the selected studies. Most articles exclusively studied invasive breast cancer (67%), followed by both invasive breast cancer and DCIS (25%). A few studies exclusively studied DCIS (7%) or were not clear whether they studied only invasive cancer, DCIS, or both (1%). The threshold for HER2 overexpression was variable between studies, with 32% using the established clinical standard [i.e., immunohistochemistry (IHC) result of 3+, or IHC result of 2+ with gene-amplification], 27% using only IHC 3+, 29% using another threshold/method, and 12% not reporting this item. Studies reporting on MRI or 18F-FDG PET more often used the clinical standard (39% and 40%, respectively) than mammography (21%) or breast ultrasound articles (25%). Whether image assessment was blinded for HER2 status was not reported in 60% of articles. In 8% of the articles, a specific breast cancer molecular subtype (predominantly ER-positive breast cancer) was excluded in their analysis (2 MRI, 2 ultrasound, and 5 mammography studies). Study results were based on BI-RADS in 64% of mammography, 36% of ultrasound, and 65% of MRI studies (excluding studies focused on non-BI-RADS imaging features such as apparent diffusion coefficient).

View this table:
Table 1.

Characteristics of studies reporting on mammography, breast ultrasound, MRI, and 18F-FDG PET imaging features in relation to HER2 overexpression in breast cancer

The results of the meta-analyses are given below. A narrative review of imaging features that were assessed in relation to HER2 overexpression in individual studies but that could not be meta-analyzed as data were too limited can be found in the Supplementary Results. Forest plots of all individual meta-analysis results are shown in Supplementary Figs. S1–S67.

Mammography imaging features in relation to HER2 overexpression in breast cancer

Study characteristics.

The 33 articles reporting on mammography features included a total of 17,745 breast cancers (40–7,281 per study), of which 2,559 were HER2-positive (14%; range, 2%–54%). The average or median reported patient age per study varied between 32 and 63 years, and tumor size between 1.0 and 4.1 cm, although the majority did not report size. Four studies investigated only DCIS, and 19 only invasive cancer. Seven studies reported to exclusively use digital mammography, four studies both analog and digital, and 14 only analog mammography. Five studies preselected patients based on receptor expression (Table 1).

Meta-analysis.

Figure 2 shows the meta-analysis results of the evaluated mammography imaging features. Five features were significantly associated with HER2 overexpression, and two reached borderline significance. The presence of microcalcifications strongly increased the chance of HER2 overexpression (pOR, 3.14; 95% CI, 2.46–4.00; P < 0.001; 23 studies), which was not dependent on whether or not a mass was associated with such microcalcifications. When microcalcifications were present, especially branching or fine linear morphology increased the chance of HER2 overexpression (pOR, 2.11; 95% CI, 1.07–4.14; P = 0.03; 6 studies), but distribution patterns of microcalcifications were not associated. Breast density also increased the chance of HER2 overexpression. In particular, BI-RADS breast density category 4 (extremely dense) showed a pOR of 1.37 (95% CI, 1.07–1.76; P = 0.01; 9 studies of which 3 were follow-up studies). Finally, a mammographic high suspicion for malignancy increased the chance of HER2 overexpression (pOR, 1.95; 95% CI, 1.22–3.10; P < 0.01; 3 studies), but circumscribed mass margins decreased the chance of HER2 overexpression by 63% (pOR, 0.37; 95% CI, 0.14–0.94; P = 0.04; 8 studies). The presence of a mass decreased the chance of HER2 overexpression with borderline significance (pOR, 0.67; 95% CI, 0.44–1.01; P = 0.06; 13 studies), and skin thickening borderline significantly increased its chance (pOR, 1.49; 95% CI, 0.95–2.34; P = 0.08; 2 studies).

Figure 2.
Figure 2.

Meta-analysis of mammography imaging features in relation to HER2 overexpression in breast cancer. The respective reference categories consist of all breast cancers within each study not belonging to the category of interest. The Egger test for publication bias can only be performed if more than two studies are available for meta-analysis. If less than two studies reported on a particular feature, no meta-analysis can be performed. Statistical significance (i.e., two-sided P < 0.05) is marked with *. Error bars depict 95% CIs.

Sources of heterogeneity in meta-analysis results.

Of above features, the presence of a mass, microcalcifications, and circumscribed margins showed marked between-study heterogeneity. Restriction of the meta-analysis to studies including invasive cancers only did not explain this heterogeneity and results were largely similar, although the finding for branching or fine linear microcalcifications did not hold anymore (pOR, 0.58; 95% CI, 0.05–7.07; P = 0.67; 3 studies), and round/oval mass shape now decreased the chance of HER2 overexpression (pOR, 0.47; 95% CI, 0.28–0.78; P < 0.01; 4 studies). After excluding studies that preselected on receptor expression, findings were predominantly similar to the overall results, but circumscribed mass margins were no longer related to HER2 overexpression (pOR, 0.66; 95% CI, 0.24–1.84; P = 0.43; 4 studies), thus partly explaining the between-study heterogeneity. In turn, indistinct mass margins became more strongly associated with HER2 overexpression (pOR, 1.48; 95% CI, 1.00–2.20; P = 0.05; 5 studies). When we considered studies using only digital or only analog mammography, findings were also similar to the overall results. Studies based on BI-RADS confirmed the results for microcalcifications, breast density, and level of suspicion, but the associations for the presence of a mass, circumscribed margins, and branching or fine linear calcification morphology became weaker (Supplementary Figs. S68–S72).

Breast ultrasound imaging features in relation to HER2 overexpression in breast cancer

Study characteristics.

The 12 articles reporting on breast ultrasound features included a total of 2,741 breast cancers (32–715 per study), of which 944 were HER2-positive (34%; range, 23%–82%). The average or median reported patient age per study varied between 32 and 57 years, and tumor size between 2.1 and 2.6 cm (half did not report size). Nine studies investigated invasive cancer only, the remainder invasive cancer in combination with DCIS. Two studies preselected on receptor expression (Table 1).

Meta-analysis.

Figure 3 shows the breast ultrasound meta-analysis results. Three features were significantly associated with HER2 overexpression, and three reached borderline significance. The presence of a mass on ultrasound decreased the chance of HER2 overexpression (pOR, 0.40; 95% CI, 0.23–0.69; P < 0.001; 4 studies), and the presence of microcalcifications increased its chance (pOR, 2.45; 95% CI, 1.27–4.72; P < 0.01; 2 studies), also when associated with a mass (pOR, 2.95; 95% CI, 2.34–3.71; P < 0.001; 7 studies). High suspicion for malignancy on ultrasound significantly increased the chance of HER2 overexpression (pOR, 5.21; 95% CI, 1.14–23.73; P = 0.03; 2 studies). Irregular shaped masses increased the chance of HER2 overexpression with borderline significance (pOR, 1.26; 95% CI, 0.99–1.60; P = 0.06; 5 studies), as did hypo-echoic echo patterns (pOR, 2.13; 95% CI, 0.91–4.99; P = 0.08; 3 studies). Circumscribed mass margins also showed a trend toward a decreased chance of HER2 overexpression (pOR, 0.66; 95% CI, 0.43–1.02; P = 0.06; 4 studies).

Figure 3.
Figure 3.

Meta-analysis of ultrasound imaging features in relation to HER2 overexpression in breast cancer. The respective reference categories consist of all breast cancers within each study not belonging to the category of interest. The Egger test for publication bias can only be performed if more than two studies are available for meta-analysis. If less than two studies reported on a particular feature, no meta-analysis can be performed. Statistical significance (i.e., two-sided P < 0.05) is marked with *. Error bars depict 95% CIs.

Sources of heterogeneity in meta-analysis results.

Restricting the meta-analysis to studies that only investigated invasive cancer did not materially change the overall results. Excluding two studies that preselected on receptor expression also did not change the results substantially, albeit the finding for circumscribed margins became less pronounced (pOR, 0.80; 95% CI, 0.50–1.27; P = 0.33; 2 studies). Studies based on BI-RADS confirmed the overall results for the presence of a mass and microcalcifications. pORs for circumscribed margin and hypo-echoic echo pattern were similar but did not reach statistical significance in this subgroup, but irregular shape became significant (pOR, 1.31; 95% CI, 1.02–1.69; P = 0.04; 4 studies; Supplementary Figs. S73–S75).

MRI features in relation to HER2 overexpression in breast cancer

Study characteristics.

The 28 articles reporting on MRI features included a total of 2,783 breast cancers (17–271 per study), of which 766 were HER2-positive (28%; range, 12%–71%). The average or median reported patient age per study varied between 46 and 60 years, and tumor size between 1.5 and 5.0 cm. Two studies investigated DCIS only and 21 studies only invasive cancers. Two studies exclusively used 3T, five both 1.5 and 3T, and 20 exclusively 1.5T systems, and one study did not provide this information. All studies used dedicated breast coils, and none preselected on molecular subtype (Table 1). Dynamic contrast enhancement (DCE)-MRI imaging protocols are shown in Supplementary Table S4.

Meta-analysis.

Figure 4 shows the DCE-MRI meta-analysis results. Overall, two DCE-MRI imaging features were significantly related to HER2 overexpression, and three showed borderline statistical significance. Fast initial kinetics showed a strong positive association with HER2 overexpression (i.e., peak enhancement at first scan: pOR, 2.60; 95% CI, 1.43–4.73; P < 0.01; 5 studies), followed by washout-type kinetic curves (pOR, 1.57; 95% CI, 1.11–2.21; P = 0.01; 12 studies). In line with the latter, persistent-type kinetic curves showed a trend toward a lower chance of HER2 overexpression (pOR, 0.60; 95% CI, 0.34–1.06; P = 0.08; 9 studies). Four other articles reported on kinetic data as well, but analyzed and presented these data in such way that we were not able to match these with descriptors used in other studies, which prevented us from including these results in the meta-analysis. Baltzer and colleagues used a semi-automated method for kinetic curve assessment and did not find an association between HER2 status and initial enhancement, washout rate, peak enhancement, or time to peak, nor between HER2 status and percentage persistent, plateau, washout, or strong enhancement (67). Kim and colleagues did not find a relation between HER2 status and initial slope, peak enhancement, time to peak enhancement, or washout ratio (76). Lee and colleagues found no relation between initial slope and HER2 status (79), and Youk and colleagues found no relation between early-phase percentage enhancement and HER2 status (92). Again following meta-analysis, irregular margins showed a trend toward increased chance of HER2 overexpression (pOR, 1.56; 95% CI, 0.99–2.47; P = 0.06; 6 studies). Furthermore, multifocality was borderline significantly related to increased HER2 overexpression (pOR, 2.45; 95% CI, 0.85–7.11; P = 0.10; 3 studies), an association that became more pronounced when restricted to mass lesions only (pOR, 5.04; 95% CI, 2.46–10.30; P < 0.001; 2 studies).

Figure 4.
Figure 4.

Meta-analysis of dynamic contrast-enhanced MRI features in relation to HER2 overexpression in breast cancer. The respective reference categories consist of all breast cancers within each study not belonging to the category of interest. The Egger test for publication bias can only be performed if more than two studies are available for meta-analysis. If less than two studies reported on a particular feature, no meta-analysis can be performed. Statistical significance (i.e., two-sided P < 0.05) is marked with *. Error bars depict 95% CIs.

DWI and MRS results are shown in the Supplementary Results.

Sources of heterogeneity in meta-analysis results.

Between-study heterogeneity in DCE-MRI imaging features was limited for above described significant and borderline significant results. Limiting the meta-analysis to invasive cancers only, studies including 3T, or 1.5T only studies, or BI-RADS–based studies did not change the findings substantially. No DCE-MRI study excluded patients based on receptor expression. Restricting the meta-analysis to three studies with sub-mm in plane resolution and smaller than 2-mm slice thickness (68, 69, 79) showed that rim enhancement also increased the chance of HER2 overexpression (pOR, 2.92; 95% CI, 1.42–6.02; P < 0.01; 2 studies; Supplementary Figs. S76–S80).

18F-FDG uptake in relation to HER2 overexpression in breast cancer

Study characteristics.

The 20 articles reporting on 18F-FDG uptake included a total of 2,027 breast cancers (12–273 per study), of which 567 were HER2-positive (28%; range, 12%–46%). The average or median reported patient age per study varied between 44 to 74 years, and tumor size between 1.6 and 5.1 cm. No study included only DCIS, five included invasive cancer and DCIS, and 15 only invasive cancer. Two studies selected on receptor expression: one study only included ER-positive breast cancer (106), and one excluded patients with HER2-positive disease who received trastuzumab within a neoadjuvant setting (97). All studies used whole-body PET or PET/CT, except one that used dedicated positron emission mammography (PEM; Table 1; ref. 107).

Meta-analysis.

After mean and SD approximation for eight studies (18, 70, 87, 94, 100, 102–104), we first meta-analyzed all 20 18F-FDG PET studies. A higher mean maximum standardized uptake value (SUVmax) in HER2-positive compared with HER2-negative breast cancer was found (pooled random effects mean difference, 0.36; 95% CI, −0.24–0.96; P = 0.24), with substantial between-study heterogeneity [I2 = 0.47; P(Q) = 0.011]. Meta-analysis of 12 studies that straightforwardly reported means and SDs did increase the observed effect slightly (pooled random effects mean difference, 0.45; 95% CI, −0.29–1.19; P = 0.23), but still with substantial heterogeneity [I2 = 0.45; P(Q) = 0.047]. A similar effect was observed when restricting the studies to the 18 without receptor expression selection [mean SUVmax difference, 0.53; 95% CI, −0.08–1.14; P = 0.09; I2 = 0.42; P(Q) = 0.034].

Meta-analysis of 10 studies reporting means and SDs and without preselection on receptor expression, likely representing the best evidence, resulted in a substantial decrease of between-study heterogeneity [I2 = 0.20; P(Q) = 0.26], and showed that HER2-positive breast cancer has a statistically significant 0.76 higher mean SUVmax than HER2-negative breast cancer (95% CI, 0.10–1.42; P = 0.025; refs. 17, 91, 93, 95, 96, 98, 99, 101, 105, 107).

Publication bias

Visual inspection of funnel plot asymmetry in combination with Egger tests generally led to a low suspicion for publication bias, albeit the number of studies was sometimes too low for proper evaluation (Supplementary Figs. S81–S147).

Discussion

We comprehensively reviewed the literature relating imaging features to HER2 overexpression in breast cancer, and pooled the results for 68 imaging features using data from 23,255 breast cancers including 4,213 HER2-positive cancers. We found 11 imaging features to be significantly related to HER2 overexpression following meta-analysis: the presence of microcalcifications on mammography or ultrasound, branching or fine linear microcalcification morphology or extremely dense breast tissue on mammography, high overall suspicion for malignancy on mammography or ultrasound, washout and fast initial kinetics on DCE-MRI, and higher 18F-FDG uptake were all associated with increased chance of HER2 overexpression, whereas the presence of a mass on ultrasound and circumscribed mass margins on mammography significantly decreased that chance.

Which exact pathophysiologic mechanisms may underlie these observations is speculative. The presence of microcalcifications indicating HER2 overexpression could be explained by the fact that DCIS, which may be a component of invasive cancer, is frequently HER2-positive and often shows microcalcifications (108–110). Alternatively, poorly differentiated tumors more often show central necrosis and rapid growth, resulting in deposition of microcalcifications along the ductal structures, and the finding could thus also be a reflection of the more aggressive nature of HER2-positive invasive cancers (109). Correlation between the presence of a mass (significant on ultrasound but also corroborated by the mammography and MRI findings) and decreased chance of HER2 overexpression is also likely a reflection of DCIS (less often presenting as a mass and more frequently HER2-positive; ref. 108). The aggressive nature of HER2-positive cancers in terms of invasiveness, cell mobility, and increased neo-angiogenesis (7) may further explain the relation with rapid kinetics and increased glucose metabolism as well as an overall high suspicion for malignancy. With regard to breast density, our results could indicate that dense breast tissue harbors a microenvironment more prone to the development of aggressive cancers, or could just merely be a reflection of younger women having both denser breasts and more frequently HER2-positive breast cancer (111, 112). The overall observation that circumscribed margins on mammography relates to less chance of HER2 overexpression may have been caused by an overrepresentation of triple-negative disease in several studies that selected patients based on receptor expression (mainly excluding ER-positive disease). Circumscribed margins, an otherwise benign feature, has been implicated as a triple-negative disease marker (113). Excluding these studies from the analysis weakened the relation between circumscribed margins and HER2 overexpression, suggesting that this feature is indeed more of a marker of triple-negative disease. A similar effect was observed for circumscribed mass margins on ultrasound.

To appreciate these findings, some strength and limitations of our meta-analysis need to be addressed. Our literature search was comprehensive and subsequent reference cross-checking of selected articles yielded only three additional indexed articles initially missed by the search. Then, the major strength of meta-analyses in general is that they increase the power to detect associations, and can identify sources of between-study heterogeneity in results. Nevertheless, a meta-analysis depends on individual study quality. Although we did not formally use a quality assessment tool, we evaluated several QUADAS-2–related study quality aspects relevant to our aims (114). For instance, we documented whether imaging assessment was performed blinded for HER2 status, which was true for 39% of articles but not reported in 60%. Most importantly, many articles ill-reported HER2 assessment methods and used variable thresholds, with only 32% of studies using the clinically established cutoff. Although this would thwart interpretation if the aim would be direct clinical application, this review focused on the biology between imaging and HER2 expression that may be visible over various HER2 threshold levels. Nevertheless, the results from this meta-analysis reflect univariable associations only, as individual studies did not adjust their results for potential confounders, such as lesion size or histologic breast cancer subtype, thus precluding solid causal inference.

The results of our meta-analysis show that imaging features relate to HER2 overexpression in breast cancer, which has several implications for future research relevant to clinical care. The diagnostic performance of breast imaging modalities may for instance improve by taking HER2 and other molecular subtypes into account when further developing rules for interpretation, resulting in “imaging signatures” specific for molecular breast cancer subtypes that are likely to outperform general rules of interpretation relating to breast cancer at large. Such imaging signatures may lead to improved detection of more aggressive breast cancer subtypes and may raise thresholds for recall and biopsy by identification of indolent disease. For example, the presence of microcalcifications with a linear/branching morphology on mammography may warrant biopsy as the chance of being consequential cancers could be high, while microcalcifications with other morphology could be actively surveyed. Also, in an era with increased use of neoadjuvant chemotherapy in which targeted therapy indication is based on pretreatment biopsies, imaging signatures predictive of HER2 overexpression may further help identify women with high risk of biopsy sampling error due to tumor heterogeneity. Finally, and perhaps most importantly, the relation between imaging features and cancer behavior may have prognostic relevance. Current clinical decisions pertaining to breast cancer treatment are based on established clinicopathologic prognostic information. Although this clinicopathologic information is pivotal for contemporary breast cancer care, the evaluation of prognostic models based on such information show that there is substantial room for improvement. For instance, the discriminative performance of such models expressed by the c-index typically is 0.70 to 0.75, with 0.5 being useless and 1.0 showing perfect discrimination (115, 116). Imaging features may thus further improve breast cancer prognostication beyond mere lesion size, and large studies are needed to establish their added value. This is especially compelling as imaging is readily available during standard diagnostic work-up for breast cancer.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Grant Support

This work was supported by a Dutch Cancer Society KWF research fellowship (UU 2010-4893 to S.G. Elias), and a René Vogels Foundation travel grant (to S.G. Elias).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Acknowledgments

The authors thank Youji He of the Southeast University (Nanjing, Jiangsu, China) for her support in interpretation of the articles written in Chinese.

Footnotes

  • Note: Supplementary data for this article are available at Cancer Epidemiology, Biomarkers & Prevention Online (http://cebp.aacrjournals.org/).

  • Received November 8, 2013.
  • Revision received January 15, 2014.
  • Accepted April 21, 2014.

References

  1. 1.↵
    1. Ferlay J,
    2. Shin HR,
    3. Bray F,
    4. Forman D,
    5. Mathers C,
    6. Parkin DM
    . GLOBOCAN 2008 v2.0, Cancer Incidence and Mortality Worldwide: IARC CancerBase No. 10. Lyon, France: International Agency for Research on Cancer; 2010 [accessed on 2013 Oct 5]. Available from: http://globocan.iarc.fr.
  2. 2.↵
    American College of Radiology. Breast imaging reporting and data system (BI-RADS). Reston, VA: American College of Radiology; 2003.
  3. 3.↵
    1. Devilee P,
    2. Tavassoli FA
    . World Health Organization: tumours of the breast and female genital organs. Oxford, United Kingdom: Oxford University Press; 2003.
  4. 4.↵
    1. Perou CM,
    2. Sorlie T,
    3. Eisen MB,
    4. van de RM,
    5. Jeffrey SS,
    6. Rees CA,
    7. et al.
    Molecular portraits of human breast tumours. Nature 2000;406:747–52.
    OpenUrlCrossRefPubMed
  5. 5.↵
    1. Curtis C,
    2. Shah SP,
    3. Chin SF,
    4. Turashvili G,
    5. Rueda OM,
    6. Dunning MJ,
    7. et al.
    The genomic and transcriptomic architecture of 2,000 breast tumours reveals novel subgroups. Nature 2012;486:346–52.
    OpenUrlCrossRefPubMed
  6. 6.↵
    1. Ferretti G,
    2. Felici A,
    3. Papaldo P,
    4. Fabi A,
    5. Cognetti F
    . HER2/neu role in breast cancer: from a prognostic foe to a predictive friend. Curr Opin Obstet Gynecol 2007;19:56–62.
    OpenUrlPubMed
  7. 7.↵
    1. Zhou BP,
    2. Hung MC
    . Dysregulation of cellular signaling by HER2/neu in breast cancer. Semin Oncol 2003;30:38–48.
    OpenUrlPubMed
  8. 8.↵
    1. Hozo SP,
    2. Djulbegovic B,
    3. Hozo I
    . Estimating the mean and variance from the median, range, and the size of a sample. BMC Med Res Methodol 2005;5:13.
    OpenUrlCrossRefPubMed
  9. 9.↵
    1. Van Belle G
    . Statistical rules of thumb. New York: Wiley-Interscience; 2002.
  10. 10.↵
    R Core Team. R: a language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2013. ISBN 3-900051-07-0, URL. Available from: http://www.R-project.org/.
  11. 11.↵
    1. Lumley T
    . rmeta: meta-analysis. R package version 2.16; 2013. Available from: http://CRAN.R-project.org/package=rmeta.
  12. 12.↵
    1. Schwarzer G
    . meta: meta-analysis with R. R package version 2.1-1; 2013. Available from: http://CRAN.R-project.org/package=meta.
  13. 13.↵
    1. Moher D,
    2. Liberati A,
    3. Tetzlaff J,
    4. Altman DG
    . Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Br Med J 2009;339:b2535.
    OpenUrlFREE Full Text
  14. 14.↵
    1. Chen M,
    2. Li JL,
    3. Song DF,
    4. Zhang Y,
    5. Li YZ,
    6. Wei LX
    . Correlation between grey scale ultrasonographic feature and estrogen receptor, progesterone receptor and human epidermal growth factor receptor-2 expression in infiltrating ductal cancer of breast. Chin J Ultrasonogr 2008;17:333–5.
    OpenUrl
  15. 15.↵
    1. Li Y,
    2. Yang L,
    3. Ding Y
    . Correlative study between ultrasonographic features and the expression of ER, PR and c-erbB2 in breast invasive ductal carcinoma. J Ultrasound in Clin Med 2008;10:452–4.
    OpenUrl
  16. 16.↵
    1. Baek HM,
    2. Chen JH,
    3. Nalcioglu O,
    4. Su MY
    . Choline as a biomarker for cell proliferation: do the results from proton MR spectroscopy show difference between HER2/neu positive and negative breast cancers? Int J Cancer 2008;123:1219–21.
    OpenUrlCrossRefPubMed
  17. 17.↵
    1. Gil-Rendo A,
    2. Martinez-Regueira F,
    3. Zornoza G,
    4. Garcia-Velloso MJ,
    5. Beorlegui C,
    6. Rodriguez-Spiteri N
    . Association between [18F]fluorodeoxyglucose uptake and prognostic parameters in breast cancer. Br J Surg 2009;96:166–70.
    OpenUrlCrossRefPubMed
  18. 18.↵
    1. Heudel P,
    2. Cimarelli S,
    3. Montella A,
    4. Bouteille C,
    5. Mognetti T
    . Value of PET-FDG in primary breast cancer based on histopathological and immunohistochemical prognostic factors. Int J Clin Oncol 2010;15:588–93.
    OpenUrlPubMed
  19. 19.↵
    1. Berriolo-Riedinger A,
    2. Touzery C,
    3. Riedinger JM,
    4. Toubeau M,
    5. Coudert B,
    6. Arnould L,
    7. et al.
    [18F]FDG-PET predicts complete pathological response of breast cancer to neoadjuvant chemotherapy. Eur J Nucl Med Mol Imaging 2007;34:1915–24.
    OpenUrlCrossRefPubMed
  20. 20.↵
    1. Buck A,
    2. Schirrmeister H,
    3. Kuhn T,
    4. Shen C,
    5. Kalker T,
    6. Kotzerke J,
    7. et al.
    FDG uptake in breast cancer: correlation with biological and clinical prognostic parameters. Eur J Nucl Med Mol Imaging 2002;29:1317–23.
    OpenUrlCrossRefPubMed
  21. 21.↵
    1. Ildefonso C,
    2. Vazquez J,
    3. Guinea O,
    4. Perez A,
    5. Fernandez A,
    6. Corte MD,
    7. et al.
    The mammographic appearance of breast carcinomas of invasive ductal type: relationship with clinicopathological parameters, biological features and prognosis. Eur J Obstet Gynecol Reprod Biol 2008;136:224–31.
    OpenUrlPubMed
  22. 22.↵
    1. Fischer U,
    2. Kopka L,
    3. Brinck U,
    4. Korabiowska M,
    5. Schauer A,
    6. Grabbe E
    . Prognostic value of contrast-enhanced MR mammography in patients with breast cancer. Eur Radiol 1997;7:1002–5.
    OpenUrlCrossRefPubMed
  23. 23.↵
    1. Shimoda W,
    2. Hayashi M,
    3. Murakami K,
    4. Oyama T,
    5. Sunagawa M
    . The relationship between FDG uptake in PET scans and biological behavior in breast cancer. Breast Cancer 2007;14:260–8.
    OpenUrlCrossRefPubMed
  24. 24.↵
    1. Specht JM,
    2. Kurland BF,
    3. Montgomery SK,
    4. Dunnwald LK,
    5. Doot RK,
    6. Gralow JR,
    7. et al.
    Tumor metabolism and blood flow as assessed by positron emission tomography varies by tumor subtype in locally advanced breast cancer. Clin Cancer Res 2010;16:2803–10.
    OpenUrlAbstract/FREE Full Text
  25. 25.↵
    1. Karamouzis MV,
    2. Likaki-Karatza E,
    3. Ravazoula P,
    4. Badra FA,
    5. Koukouras D,
    6. Tzorakoleftherakis E,
    7. et al.
    Non-palpable breast carcinomas: correlation of mammographically detected malignant-appearing microcalcifications and molecular prognostic factors. Int J Cancer 2002;102:86–90.
    OpenUrlCrossRefPubMed
  26. 26.↵
    1. Kumar R,
    2. Chauhan A,
    3. Zhuang H,
    4. Chandra P,
    5. Schnall M,
    6. Alavi A
    . Clinicopathologic factors associated with false negative FDG-PET in primary breast cancer. Breast Cancer Res Treat 2006;98:267–74.
    OpenUrlCrossRefPubMed
  27. 27.↵
    1. Ueda S,
    2. Kondoh N,
    3. Tsuda H,
    4. Yamamoto S,
    5. Asakawa H,
    6. Fukatsu K,
    7. et al.
    Expression of centromere protein F (CENP-F) associated with higher FDG uptake on PET/CT, detected by cDNA microarray, predicts high-risk patients with primary breast cancer. BMC Cancer 2008;8:384.
    OpenUrlCrossRefPubMed
  28. 28.↵
    1. Agrawal AK,
    2. Jelen M,
    3. Rudnicki J,
    4. Grzebieniak Z,
    5. Zukrowski P,
    6. Nienartowicz E
    . Molecular markers (c-erbB-2, p53) in breast cancer. Folia Histochem Cytobiol 2008;46:449–55.
    OpenUrlCrossRefPubMed
  29. 29.↵
    1. Aiello EJ,
    2. Buist DSM,
    3. White E,
    4. Porter PL
    . Association between mammographic breast density and breast cancer tumor characteristics. Cancer Epidemiol Biomarkers Prev 2005;14:662–8.
    OpenUrlAbstract/FREE Full Text
  30. 30.↵
    1. Arora N,
    2. King TA,
    3. Jacks LM,
    4. Stempel MM,
    5. Patil S,
    6. Morris E,
    7. et al.
    Impact of breast density on the presenting features of malignancy. Ann Surg Oncol 2010;17(Suppl 3):211–8.
    OpenUrl
  31. 31.↵
    1. Badra FA,
    2. Karamouzis MV,
    3. Ravazoula P,
    4. Likaki-Karatza E,
    5. Tzorakoleftherakis E,
    6. Koukouras D,
    7. et al.
    Non-palpable breast carcinomas: correlation of mammographically detected malignant-appearing microcalcifications and epidermal growth factor receptor (EGFR) family expression. Cancer Lett 2006;244:34–41.
    OpenUrlPubMed
  32. 32.↵
    1. Cui CX,
    2. Lin Q,
    3. Yang Q,
    4. Zhang CY,
    5. Wang SH,
    6. Yu HL,
    7. et al.
    Comparative study on mammography between triple negative and triple positive breast cancer. Chin J Radiol 2012;46:420–4.
    OpenUrl
  33. 33.↵
    1. Di Nubila B,
    2. Cassano E,
    3. Urban LA,
    4. Fedele P,
    5. Abbate F,
    6. Maisonneuve P,
    7. et al.
    Radiological features and pathological–biological correlations in 348 women with breast cancer under 35 years old. Breast 2006;15:744–53.
    OpenUrlCrossRefPubMed
  34. 34.↵
    1. Enache DE,
    2. Georgescu CV,
    3. Patrana N
    . Negative estrogen-receptor invasive breast carcinoma: mammographic aspects, correlations with HER2/neu oncoprotein status. Rom J Morphol Embryol 2012;53:755–62.
    OpenUrlPubMed
  35. 35.↵
    1. Evans AJ,
    2. Pinder SE,
    3. Ellis IO,
    4. Sibbering DM,
    5. Elston CW,
    6. Poller DN,
    7. et al.
    Correlations between the mammographic features of ductal carcinoma in situ (DCIS) and C-erbB-2 oncogene expression. Nottingham Breast Team. Clin Radiol 1994;49:559–62.
    OpenUrlCrossRefPubMed
  36. 36.↵
    1. Fasching PA,
    2. Heusinger K,
    3. Loehberg CR,
    4. Wenkel E,
    5. Lux MP,
    6. Schrauder M,
    7. et al.
    Influence of mammographic density on the diagnostic accuracy of tumor size assessment and association with breast cancer tumor characteristics. Eur J Radiol 2006;60:398–404.
    OpenUrlCrossRefPubMed
  37. 37.↵
    1. Gajdos C,
    2. Tartter PI,
    3. Bleiweiss IJ,
    4. Hermann G,
    5. de Csepel J,
    6. Estabrook A,
    7. et al.
    Mammographic appearance of nonpalpable breast cancer reflects pathologic characteristics. Ann Surg 2002;235:246–51.
    OpenUrlCrossRefPubMed
  38. 38.↵
    1. Gu YJ,
    2. Xiao Q,
    3. Yang WT,
    4. Zheng XJ,
    5. Gu RF
    . The X-ray features of breast ductal carcinoma in situ and its small invasive foci and correlation between mammographic features and prognostic biologic factors. Chin J Radiol 2007;41:623–8.
    OpenUrl
  39. 39.↵
    1. Jiang L,
    2. Ma T,
    3. Moran MS,
    4. Kong X,
    5. Li X,
    6. Haffty BG,
    7. et al.
    Mammographic features are associated with clinicopathological characteristics in invasive breast cancer. Anticancer Res 2011;31:2327–34.
    OpenUrlAbstract/FREE Full Text
  40. 40.↵
    1. Kim JH,
    2. Ko ES,
    3. Kim dY,
    4. Han H,
    5. Sohn JH,
    6. Choe dH
    . Noncalcified ductal carcinoma in situ: imaging and histologic findings in 36 tumors. J Ultrasound Med 2009;28:903–10.
    OpenUrlAbstract/FREE Full Text
  41. 41.↵
    1. Ko ES,
    2. Lee BH,
    3. Kim HA,
    4. Noh WC,
    5. Kim MS,
    6. Lee SA
    . Triple-negative breast cancer: correlation between imaging and pathological findings. Eur Radiol 2010;20:1111–7.
    OpenUrlCrossRefPubMed
  42. 42.↵
    1. Kuo YJ,
    2. Ho DMT,
    3. Tsai YF,
    4. Hsu CY
    . Invasive ductal carcinoma arising in phyllodes tumor with isolated tumor cells in sentinel lymph node. J Chin Med Assoc 2010;73:602–4.
    OpenUrlPubMed
  43. 43.↵
    1. Li R,
    2. Chen Y
    . Mammographic features and clinical characteristics of different molecular subtypes of infiltrative ductal carcinoma. Chin J Med Imaging Technol 2011;27:565–8.
    OpenUrl
  44. 44.↵
    1. Ma H,
    2. Luo J,
    3. Press MF,
    4. Wang Y,
    5. Bernstein L,
    6. Ursin G
    . Is there a difference in the association between percent mammographic density and subtypes of breast cancer? Luminal A and triple-negative breast cancer. Cancer Epidemiol Biomarkers Prev 2009;18:479–85.
    OpenUrlAbstract/FREE Full Text
  45. 45.↵
    1. Mansson E,
    2. Bergkvist L,
    3. Christenson G,
    4. Persson C,
    5. Warnberg F
    . Mammographic casting-type calcifications is not a prognostic factor in unifocal small invasive breast cancer: a population-based retrospective cohort study. J Surg Oncol 2009;100:670–4.
    OpenUrlPubMed
  46. 46.↵
    1. Mun HS,
    2. Shin HJ,
    3. Kim HH,
    4. Cha JH,
    5. Kim H
    . Screening-detected calcified and non-calcified ductal carcinoma in situ: differences in the imaging and histopathological features. Clin Radiol 2013;68:e27–35.
    OpenUrlCrossRefPubMed
  47. 47.↵
    1. Palka I,
    2. Ormandi K,
    3. Gaal S,
    4. Boda K,
    5. Kahan Z
    . Casting-type calcifications on the mammogram suggest a higher probability of early relapse and death among high-risk breast cancer patients. Acta Oncol 2007;46:1178–83.
    OpenUrlPubMed
  48. 48.↵
    1. Palka I,
    2. Kelemen G,
    3. Ormandi K,
    4. Lazar G,
    5. Nyari T,
    6. Thurzo L,
    7. et al.
    Tumor characteristics in screen-detected and symptomatic breast cancers. Pathol Oncol Res 2008;14:161–7.
    OpenUrlCrossRefPubMed
  49. 49.↵
    1. Phipps AI,
    2. Buist DS,
    3. Malone KE,
    4. Barlow WE,
    5. Porter PL,
    6. Kerlikowske K,
    7. et al.
    Breast density, body mass index, and risk of tumor marker-defined subtypes of breast cancer. Ann Epidemiol 2012;22:340–8.
    OpenUrlPubMed
  50. 50.↵
    1. Pollan M,
    2. Ascunce N,
    3. Ederra M,
    4. Murillo A,
    5. Erdozain N,
    6. es-Martinez JE,
    7. et al.
    Mammographic density and risk of breast cancer according to tumor characteristics and mode of detection: a Spanish population-based case–control study. Breast Cancer Res 2013;15:R9.
    OpenUrlPubMed
  51. 51.↵
    1. Seo BK,
    2. Pisano ED,
    3. Kuzimak CM,
    4. Koomen M,
    5. Pavic D,
    6. Lee Y,
    7. et al.
    Correlation of HER-2/neu overexpression with mammography and age distribution in primary breast carcinomas. Acad Radiol 2006;13:1211–8.
    OpenUrlCrossRefPubMed
  52. 52.↵
    1. Shin HJ,
    2. Kim HH,
    3. Huh MO,
    4. Kim MJ,
    5. Yi A,
    6. Kim H,
    7. 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.
    OpenUrlAbstract/FREE Full Text
  53. 53.↵
    1. Sun QH,
    2. Fu RZ,
    3. Guo F,
    4. Ren LJ
    . Correlation between feature-calcification of mamnographic X-ray and expression of c-erbB-2 protein in breast cancer. Chin J Cancer Prev Treat 2008;15:1481–2.
    OpenUrl
  54. 54.↵
    1. Taneja S,
    2. Evans AJ,
    3. Rakha EA,
    4. Green AR,
    5. Ball G,
    6. Ellis IO
    . The mammographic correlations of a new immunohistochemical classification of invasive breast cancer. Clin Radiol 2008;63:1228–35.
    OpenUrlPubMed
  55. 55.↵
    1. Wang X,
    2. Chao L,
    3. Chen L,
    4. Tian B,
    5. Ma G,
    6. Zang Y,
    7. et al.
    Correlation of mammographic calcifications with Her-2/neu overexpression in primary breast carcinomas. J Digit Imaging 2008;21:170–6.
    OpenUrlCrossRefPubMed
  56. 56.↵
    1. Wang Y,
    2. Ikeda DM,
    3. Narasimhan B,
    4. Longacre TA,
    5. Bleicher RJ,
    6. Pal S,
    7. et al.
    Estrogen receptor–negative invasive breast cancer: imaging features of tumors with and without human epidermal growth factor receptor type 2 overexpression. Radiology 2008;246:367–75.
    OpenUrlCrossRefPubMed
  57. 57.↵
    1. Wang YX,
    2. Xu XJ
    . Correlation study on mammographic features, pathology and molecular biology of breast cancer. Chin J Med Imaging Technol 2008;24:60–3.
    OpenUrl
  58. 58.↵
    1. Wen SY,
    2. Han YW,
    3. Ma XM,
    4. Zhang J,
    5. Cui WJ,
    6. Cao XC,
    7. et al.
    [Mammographic and pathological features of triple-negative breast cancer]. Zhonghua Zhong Liu Za Zhi 2012;34:291–5.
    OpenUrlPubMed
  59. 59.↵
    1. Yaghjyan L,
    2. Colditz GA,
    3. Collins LC,
    4. Schnitt SJ,
    5. Rosner B,
    6. Vachon C,
    7. et al.
    Mammographic breast density and subsequent risk of breast cancer in postmenopausal women according to tumor characteristics. J Natl Cancer Inst 2011;103:1179–89.
    OpenUrlAbstract/FREE Full Text
  60. 60.↵
    1. Yang WT,
    2. Dryden M,
    3. Broglio K,
    4. Gilcrease M,
    5. Dawood S,
    6. Dempsey PJ,
    7. et al.
    Mammographic features of triple receptor-negative primary breast cancers in young premenopausal women. Breast Cancer Res Treat 2008;111:405–10.
    OpenUrlCrossRefPubMed
  61. 61.↵
    1. Au-Yong IT,
    2. Evans AJ,
    3. Taneja S,
    4. Rakha EA,
    5. Green AR,
    6. Paish C,
    7. et al.
    Sonographic correlations with the new molecular classification of invasive breast cancer. Eur Radiol 2009;19:2342–8.
    OpenUrlCrossRefPubMed
  62. 62.↵
    1. Chen ST,
    2. Kuo SJ,
    3. Wu HK,
    4. Chen LS,
    5. Chen DR
    . Power Doppler breast ultrasound: association of vascularization and ER/c-erbB-2 co-expression in invasive breast carcinoma. Breast Cancer 2011;20:152–8.
    OpenUrlPubMed
  63. 63.↵
    1. Kim SH,
    2. Seo BK,
    3. Lee J,
    4. Kim SJ,
    5. Cho KR,
    6. Lee KY,
    7. et al.
    Correlation of ultrasound findings with histology, tumor grade, and biological markers in breast cancer. Acta Oncol 2008;47:1531–8.
    OpenUrlCrossRefPubMed
  64. 64.↵
    1. Pang CX,
    2. Li ZX,
    3. Ding XM,
    4. Zhang L,
    5. Wei KL,
    6. Huang KQ
    . Infiltrating ductal cancer of breast: correlation between ultrasonographic features and ER, C-erbB-2 expression level. Chin J Med Imaging Technol 2009;25:449–52.
    OpenUrl
  65. 65.↵
    1. Wan CF,
    2. Du J,
    3. Fang H,
    4. Li FH,
    5. Zhu JS,
    6. Liu Q
    . Enhancement patterns and parameters of breast cancers at contrast-enhanced US: correlation with prognostic factors. Radiology 2012;262:450–9.
    OpenUrlCrossRefPubMed
  66. 66.↵
    1. Agrawal G,
    2. Chen JH,
    3. Baek HM,
    4. Hsiang D,
    5. Mehta RS,
    6. Nalcioglu O,
    7. et al.
    MRI features of breast cancer: a correlation study with HER-2 receptor. Ann Oncol 2007;18:1903–4.
    OpenUrlFREE Full Text
  67. 67.↵
    1. Baltzer PA,
    2. Vag T,
    3. Dietzel M,
    4. Beger S,
    5. Freiberg C,
    6. Gajda M,
    7. et al.
    Computer-aided interpretation of dynamic magnetic resonance imaging reflects histopathology of invasive breast cancer. Eur Radiol 2010;20:1563–71.
    OpenUrlCrossRefPubMed
  68. 68.↵
    1. Chang YW,
    2. Kwon KH,
    3. Choi DL,
    4. Lee DW,
    5. Lee MH,
    6. Lee HK,
    7. et al.
    Magnetic resonance imaging of breast cancer and correlation with prognostic factors. Acta Radiol 2009;50:990–8.
    OpenUrlAbstract/FREE Full Text
  69. 69.↵
    1. Chen JH,
    2. Bahri S,
    3. Mehta RS,
    4. Kuzucan A,
    5. Yu HJ,
    6. Carpenter PM,
    7. et al.
    Breast cancer: evaluation of response to neoadjuvant chemotherapy with 3.0-T MR imaging. Radiology 2011;261:735–43.
    OpenUrlCrossRefPubMed
  70. 70.↵
    1. Choi BB,
    2. Kim SH,
    3. Kang BJ,
    4. Lee JH,
    5. Song BJ,
    6. Jeong SH,
    7. et al.
    Diffusion-weighted imaging and FDG PET/CT: predicting the prognoses with apparent diffusion coefficient values and maximum standardized uptake values in patients with invasive ductal carcinoma. World J Surg Oncol 2012;10:126.
    OpenUrlCrossRefPubMed
  71. 71.↵
    1. Costantini M,
    2. Belli P,
    3. Distefano D,
    4. Bufi E,
    5. Matteo MD,
    6. Rinaldi P,
    7. et al.
    Magnetic resonance imaging features in triple-negative breast cancer: comparison with luminal and HER2-overexpressing tumors. Clin Breast Cancer 2012;12:331–9.
    OpenUrlPubMed
  72. 72.↵
    1. Fernandez-Guinea O,
    2. Andicoechea A,
    3. Gonzalez LO,
    4. Gonzalez-Reyes S,
    5. Merino AM,
    6. Hernandez LC,
    7. et al.
    Relationship between morphological features and kinetic patterns of enhancement of the dynamic breast magnetic resonance imaging and clinico-pathological and biological factors in invasive breast cancer. BMC Cancer 2010;10:8.
    OpenUrlCrossRefPubMed
  73. 73.↵
    1. Girardi V,
    2. Carbognin G,
    3. Camera L,
    4. Tonegutti M,
    5. Bonetti F,
    6. Manfrin E,
    7. et al.
    Fischer's score criteria correlating with histopathological prognostic factors in invasive breast cancer. Radiol Med 2010;115:421–33.
    OpenUrlPubMed
  74. 74.↵
    1. Gomez-Raposo C,
    2. Andreu M,
    3. Suarez-Garcia I,
    4. Esteban MI,
    5. Carballo M,
    6. Sereno-Moyano MF,
    7. et al.
    Relevance of breast cancer subtypes for magnetic resonance imaging response monitoring during neoadjuvant chemotherapy. Clin Transl Oncol 2012;14:486–8.
    OpenUrlPubMed
  75. 75.↵
    1. Jeh SK,
    2. Kim SH,
    3. Kim HS,
    4. Kang BJ,
    5. Jeong SH,
    6. Yim HW,
    7. et al.
    Correlation of the apparent diffusion coefficient value and dynamic magnetic resonance imaging findings with prognostic factors in invasive ductal carcinoma. J Magn Reson Imaging 2011;33:102–9.
    OpenUrlCrossRefPubMed
  76. 76.↵
    1. Kim JA,
    2. Son EJ,
    3. Youk JH,
    4. Kim EK,
    5. Kim MJ,
    6. Kwak JY,
    7. et al.
    MRI findings of pure ductal carcinoma in situ: kinetic characteristics compared according to lesion type and histopathologic factors. Am J Roentgenol 2011;196:1450–6.
    OpenUrlCrossRefPubMed
  77. 77.↵
    1. Kim SH,
    2. Cha ES,
    3. Kim HS,
    4. Kang BJ,
    5. Choi JJ,
    6. Jung JH,
    7. et al.
    Diffusion-weighted imaging of breast cancer: correlation of the apparent diffusion coefficient value with prognostic factors. J Magn Reson Imaging 2009;30:615–20.
    OpenUrlCrossRefPubMed
  78. 78.↵
    1. Koo HR,
    2. Cho N,
    3. Song IC,
    4. Kim H,
    5. Chang JM,
    6. Yi A,
    7. et al.
    Correlation of perfusion parameters on dynamic contrast-enhanced MRI with prognostic factors and subtypes of breast cancers. J Magn Reson Imaging 2012;36:145–51.
    OpenUrlPubMed
  79. 79.↵
    1. Lee SH,
    2. Cho N,
    3. Kim SJ,
    4. Cha JH,
    5. Cho KS,
    6. Ko ES,
    7. et al.
    Correlation between high resolution dynamic MR features and prognostic factors in breast cancer. Korean J Radiol 2008;9:10–8.
    OpenUrlCrossRefPubMed
  80. 80.↵
    1. Liu H,
    2. Peng W
    . MRI morphological classification of ductal carcinoma in situ (DCIS) correlating with different biological behavior. Eur J Radiol 2011;81:214–7.
    OpenUrlPubMed
  81. 81.↵
    1. Loo CE,
    2. Straver ME,
    3. Rodenhuis S,
    4. Muller SH,
    5. Wesseling J,
    6. Vrancken Peeters MJ,
    7. et al.
    Magnetic resonance imaging response monitoring of breast cancer during neoadjuvant chemotherapy: relevance of breast cancer subtype. J Clin Oncol 2011;29:660–6.
    OpenUrlAbstract/FREE Full Text
  82. 82.↵
    1. Lu H,
    2. Xu YL,
    3. Zhang SP,
    4. Lang RG,
    5. Zee CS,
    6. Liu PF,
    7. et al.
    Breast magnetic resonance imaging in patients with occult breast carcinoma: evaluation on feasibility and correlation with histopathological findings. Chin Med J 2011;124:1790–5.
    OpenUrlPubMed
  83. 83.↵
    1. Makkat S,
    2. Luypaert R,
    3. Stadnik T,
    4. Bourgain C,
    5. Sourbron S,
    6. Dujardin M,
    7. et al.
    Deconvolution-based dynamic contrast-enhanced MR imaging of breast tumors: correlation of tumor blood flow with human epidermal growth factor receptor 2 status and clinicopathologic findings—preliminary results. Radiology 2008;249:471–82.
    OpenUrlCrossRefPubMed
  84. 84.↵
    1. Marcos de Paz LM,
    2. Tejerina BA,
    3. Arranz Merino ML,
    4. Calvo de Juan V
    . Breast MR imaging changes after neoadjuvant chemotherapy: correlation with molecular subtypes. Radiologia 2011;54:442–8.
    OpenUrlPubMed
  85. 85.↵
    1. Martincich L,
    2. Deantoni V,
    3. Bertotto I,
    4. Redana S,
    5. Kubatzki F,
    6. Sarotto I,
    7. et al.
    Correlations between diffusion-weighted imaging and breast cancer biomarkers. Eur Radiol 2012;22:1519–28.
    OpenUrlCrossRefPubMed
  86. 86.↵
    1. Montemurro F,
    2. Martincich L,
    3. Sarotto I,
    4. Bertotto I,
    5. Ponzone R,
    6. Cellini L,
    7. et al.
    Relationship between DCE-MRI morphological and functional features and histopathological characteristics of breast cancer. Eur Radiol 2007;17:1490–7.
    OpenUrlCrossRefPubMed
  87. 87.↵
    1. Nakajo M,
    2. Kajiya Y,
    3. Kaneko T,
    4. Kaneko Y,
    5. Takasaki T,
    6. Tani A,
    7. et al.
    FDG PET/CT and diffusion-weighted imaging for breast cancer: prognostic value of maximum standardized uptake values and apparent diffusion coefficient values of the primary lesion. Eur J Nucl Med Mol Imaging 2010;37:2011–20.
    OpenUrlCrossRefPubMed
  88. 88.↵
    1. Sah RG,
    2. Sharma U,
    3. Parshad R,
    4. Seenu V,
    5. Mathur SR,
    6. Jagannathan NR
    . Association of estrogen receptor, progesterone receptor, and human epidermal growth factor receptor 2 status with total choline concentration and tumor volume in breast cancer patients: an MRI and in vivo proton MRS study. Magn Reson Med 2012;68:1039–47.
    OpenUrlCrossRefPubMed
  89. 89.↵
    1. Shin HJ,
    2. Baek HM,
    3. Cha JH,
    4. Kim HH
    . Evaluation of breast cancer using proton MR spectroscopy: total choline peak integral and signal-to-noise ratio as prognostic indicators. Am J Roentgenol 2012;198:W488–97.
    OpenUrlCrossRefPubMed
  90. 90.↵
    1. Szabo BK,
    2. Aspelin P,
    3. Kristoffersen WM,
    4. Tot T,
    5. Bone B
    . Invasive breast cancer: correlation of dynamic MR features with prognostic factors. Eur Radiol 2003;13:2425–35.
    OpenUrlCrossRefPubMed
  91. 91.↵
    1. Tozaki M,
    2. Hoshi K
    . 1H MR spectroscopy of invasive ductal carcinoma: correlations with FDG PET and histologic prognostic factors. Am J Roentgenol 2010;194:1384–90.
    OpenUrlCrossRefPubMed
  92. 92.↵
    1. Youk JH,
    2. Son EJ,
    3. Chung J,
    4. Kim JA,
    5. Kim EK
    . Triple-negative invasive breast cancer on dynamic contrast-enhanced and diffusion-weighted MR imaging: comparison with other breast cancer subtypes. Eur Radiol 2012;22:1724–34.
    OpenUrlCrossRefPubMed
  93. 93.↵
    1. Garcia Vicente AM,
    2. Castrejon AS,
    3. Relea CF,
    4. Munoz AP,
    5. Leon Martin AA,
    6. Lopez-Muniz IC,
    7. et al.
    18F-FDG retention index and biologic prognostic parameters in breast cancer. Clin Nucl Med 2012;37:460–6.
    OpenUrlPubMed
  94. 94.↵
    1. Groheux D,
    2. Giacchetti S,
    3. Moretti JL,
    4. Porcher R,
    5. Espie M,
    6. Lehmann-Che J,
    7. et al.
    Correlation of high 18F-FDG uptake to clinical, pathological and biological prognostic factors in breast cancer. Eur J Nucl Med Mol Imaging 2011;38:426–35.
    OpenUrlCrossRefPubMed
  95. 95.↵
    1. Humbert O,
    2. Berriolo-Riedinger A,
    3. Riedinger JM,
    4. Coudert B,
    5. Arnould L,
    6. Cochet A,
    7. et al.
    Changes in 18F-FDG tumor metabolism after a first course of neoadjuvant chemotherapy in breast cancer: influence of tumor subtypes. Ann Oncol 2012;23:2572–7.
    OpenUrlAbstract/FREE Full Text
  96. 96.↵
    1. Ikenaga N,
    2. Otomo N,
    3. Toyofuku A,
    4. Ueda Y,
    5. Toyoda K,
    6. Hayashi T,
    7. et al.
    Standardized uptake values for breast carcinomas assessed by fluorodeoxyglucose-positron emission tomography correlate with prognostic factors. Am Surg 2007;73:1151–7.
    OpenUrlPubMed
  97. 97.↵
    1. Jin S,
    2. Kim SB,
    3. Ahn JH,
    4. Jung KH,
    5. Ahn SH,
    6. Son BH,
    7. et al.
    18F-fluorodeoxyglucose uptake predicts pathological complete response after neoadjuvant chemotherapy for breast cancer: a retrospective cohort study. J Surg Oncol 2013;107:180–7.
    OpenUrlPubMed
  98. 98.↵
    1. Keam B,
    2. Im SA,
    3. Koh Y,
    4. Han SW,
    5. Oh DY,
    6. Cho N,
    7. et al.
    Early metabolic response using FDG PET/CT and molecular phenotypes of breast cancer treated with neoadjuvant chemotherapy. BMC Cancer 2011;11:254.
    OpenUrlCrossRefPubMed
  99. 99.↵
    1. Kim BS,
    2. Sung SH
    . Usefulness of 18F-FDG uptake with clinicopathologic and immunohistochemical prognostic factors in breast cancer. Ann Nucl Med 2012;26:175–83.
    OpenUrlCrossRefPubMed
  100. 100.↵
    1. Koolen BB,
    2. Vrancken Peeters MJTF,
    3. Wesseling J,
    4. Lips EH,
    5. Vogel WV,
    6. Aukema TS,
    7. et al.
    Association of primary tumour FDG uptake with clinical, histopathological and molecular characteristics in breast cancer patients scheduled for neoadjuvant chemotherapy. Eur J Nucl Med Mol Imaging 2012;39:1830–8.
    OpenUrlPubMed
  101. 101.↵
    1. Mavi A,
    2. Cermik TF,
    3. Urhan M,
    4. Puskulcu H,
    5. Basu S,
    6. Yu JQ,
    7. et al.
    The effects of estrogen, progesterone, and C-erbB-2 receptor states on 18F-FDG uptake of primary breast cancer lesions. J Nucl Med 2007;48:1266–72.
    OpenUrlAbstract/FREE Full Text
  102. 102.↵
    1. Osborne JR,
    2. Port E,
    3. Gonen M,
    4. Doane A,
    5. Yeung H,
    6. Gerald W,
    7. et al.
    18F-FDG PET of locally invasive breast cancer and association of estrogen receptor status with standardized uptake value: microarray and immunohistochemical analysis. J Nucl Med 2010;51:543–50.
    OpenUrlAbstract/FREE Full Text
  103. 103.↵
    1. Sanli Y,
    2. Kuyumcu S,
    3. Ozkan ZG,
    4. Isik G,
    5. Karanlik H,
    6. Guzelbey B,
    7. et al.
    Increased FDG uptake in breast cancer is associated with prognostic factors. Ann Nucl Med 2012;26:345–50.
    OpenUrlPubMed
  104. 104.↵
    1. Straver ME,
    2. Aukema TS,
    3. Olmos RA,
    4. Rutgers EJ,
    5. Gilhuijs KG,
    6. Schot ME,
    7. et al.
    Feasibility of FDG PET/CT to monitor the response of axillary lymph node metastases to neoadjuvant chemotherapy in breast cancer patients. Eur J Nucl Med Mol Imaging 2010;37:1069–76.
    OpenUrlCrossRefPubMed
  105. 105.↵
    1. Ueda S,
    2. Tsuda H,
    3. Asakawa H,
    4. Shigekawa T,
    5. Fukatsu K,
    6. Kondo N,
    7. et al.
    Clinicopathological and prognostic relevance of uptake level using 18F-fluorodeoxyglucose positron emission tomography/computed tomography fusion imaging (18F-FDG PET/CT) in primary breast cancer. Jpn J Clin Oncol 2008;38:250–8.
    OpenUrlAbstract/FREE Full Text
  106. 106.↵
    1. Ueda S,
    2. Tsuda H,
    3. Saeki T,
    4. Omata J,
    5. Osaki A,
    6. Shigekawa T,
    7. et al.
    Early metabolic response to neoadjuvant letrozole, measured by FDG PET/CT, is correlated with a decrease in the Ki67 labeling index in patients with hormone receptor–positive primary breast cancer: a pilot study. Breast Cancer 2011;18:299–308.
    OpenUrlPubMed
  107. 107.↵
    1. Wang CL,
    2. MacDonald LR,
    3. Rogers JV,
    4. Aravkin A,
    5. Haseley DR,
    6. Beatty JD
    . Positron emission mammography: correlation of estrogen receptor, progesterone receptor, and human epidermal growth factor receptor 2 status and 18F-FDG. Am J Roentgenol 2011;197:W247–55.
    OpenUrlCrossRefPubMed
  108. 108.↵
    1. Yamada T,
    2. Mori N,
    3. Watanabe M,
    4. Kimijima I,
    5. Okumoto T,
    6. Seiji K,
    7. et al.
    Radiologic–pathologic correlation of ductal carcinoma in situ . Radiographics 2010;30:1183–98.
    OpenUrlCrossRefPubMed
  109. 109.↵
    1. Tse GM,
    2. Tan PH,
    3. Pang AL,
    4. Tang AP,
    5. Cheung HS
    . Calcification in breast lesions: pathologists' perspective. J Clin Pathol 2008;61:145–51.
    OpenUrlAbstract/FREE Full Text
  110. 110.↵
    1. Allred DC,
    2. Clark GM,
    3. Molina R,
    4. Tandon AK,
    5. Schnitt SJ,
    6. Gilchrist KW,
    7. et al.
    Overexpression of HER-2/neu and its relationship with other prognostic factors change during the progression of in situ to invasive breast cancer. Hum Pathol 1992;23:974–9.
    OpenUrlCrossRefPubMed
  111. 111.↵
    1. Anders CK,
    2. Hsu DS,
    3. Broadwater G,
    4. Acharya CR,
    5. Foekens JA,
    6. Zhang Y,
    7. et al.
    Young age at diagnosis correlates with worse prognosis and defines a subset of breast cancers with shared patterns of gene expression. J Clin Oncol 2008;26:3324–30.
    OpenUrlAbstract/FREE Full Text
  112. 112.↵
    1. Boyd NF,
    2. Rommens JM,
    3. Vogt K,
    4. Lee V,
    5. Hopper JL,
    6. Yaffe MJ,
    7. et al.
    Mammographic breast density as an intermediate phenotype for breast cancer. Lancet Oncol 2005;6:798–808.
    OpenUrlCrossRefPubMed
  113. 113.↵
    1. Dogan BE,
    2. Turnbull LW
    . Imaging of triple-negative breast cancer. Ann Oncol 2012;23(Suppl 6):vi23–9.
    OpenUrlAbstract/FREE Full Text
  114. 114.↵
    1. Whiting PF,
    2. Rutjes AW,
    3. Westwood ME,
    4. Mallett S,
    5. Deeks JJ,
    6. Reitsma JB,
    7. et al.
    QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies. Ann Intern Med 2011;155:529–36.
    OpenUrlCrossRefPubMed
  115. 115.↵
    1. Campbell HE,
    2. Gray AM,
    3. Harris AL,
    4. Briggs AH,
    5. Taylor MA
    . Estimation and external validation of a new prognostic model for predicting recurrence-free survival for early breast cancer patients in the UK. Br J Cancer 2010;103:776–86.
    OpenUrlCrossRefPubMed
  116. 116.↵
    1. Mook S,
    2. Schmidt MK,
    3. Rutgers EJ,
    4. van d V,
    5. Visser O,
    6. Rutgers SM,
    7. et al.
    Calibration and discriminatory accuracy of prognosis calculation for breast cancer with the online Adjuvant! program: a hospital-based retrospective cohort study. Lancet Oncol 2009;10:1070–6.
    OpenUrlCrossRefPubMed
View Abstract
Back to top
Cancer Epidemiology Biomarkers & Prevention: 23 (8)
August 2014
Volume 23, Issue 8

Sign up for alerts

View this article with LENS

Open full page PDF
Article Alerts
Email Article
Citation Tools
Imaging Features of HER2 Overexpression in Breast Cancer: A Systematic Review and Meta-analysis
Sjoerd G. Elias, Arthur Adams, Dorota J. Wisner, Laura J. Esserman, Laura J. van't Veer, Willem P.Th.M. Mali, Kenneth G.A. Gilhuijs and Nola M. Hylton
Cancer Epidemiol Biomarkers Prev August 1 2014 (23) (8) 1464-1483; DOI: 10.1158/1055-9965.EPI-13-1170
Digg logo Reddit logo Twitter logo Facebook logo Google logo Mendeley logo

Jump to section

Advertisement