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.

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.

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.

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.

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.

Close modal

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).

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).

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

PatientsBreast cancersHER2 assessmentImaging acquisition and results
HistologycPositiveReaders
Ref., country, recruitment periodRecruitmentNAge in years (range/SD)NSize in cm (range/SD)IDCElseDCISAssaydCutoffN (%)Imaging system (manufacturer)Imaging parametersNBlindedImaging features described in reviewe,f,g,h
Mammography             Analog/Digital    
Agrawal et al. (28), Poland, 1992–2001 University hospital 184 57a (32–75) 184 NS (NS)    IHC Dakopatts (K-353) NS 100 (54) NS Analog NS NS Level of suspicion 
Aiello et al. (29), USA, 1988–1995 Breast cancer screening program 461 NS (NS) 461 NS (0.1–10) NOS  IHC NOS Other 79 (17) NS Analog NS BrDe 
Arora et al. (30), USA, 2005–2007 Cancer center; no prior breast cancer 1,323 56b (27–91) 1,323 1.5b (0.1–11)    IHC NOS NS 170 (13) NS Analog, digital NS NS BrDe 
Badra et al. (31), Greece, 1994–2004 University hospital; nonpalpable screen-detected 75 56b (35–74) 75 NS (NS)    IHC BioGenex (CB11)  34 (45) NS NS NS Cal 
Cui et al. (32), China, 2009–2011 University hospital; triple-negative and triple-positive only 161 51a (27–87) 161 2.4a (NS)    IHC NOS  44 (27) Senographe 800T (GE) Analog Yes BrDe; Mas (Mar; Sha; Cal); Cal 
Di Nubila et al. (33), Italy, 2000–2002 Cancer center; ≤ 35 y, no prior breast cancer 232 32a (22–35) 232 2.3a (NS)    IHC Triton (TAB250)  64 (28) Senographe 2000D; Senographe DMR (GE) Analog, digital NS NS Level of suspicion; BrDe; Cal 
Enache et al. (34), Romania, 2010–2011 General hospital; only ER 40 51a (25–88) 40 4.1a (0.6–10.7)    IHC Dako (polyclonal); CISH Invitrogen  11 (28) NS NS NS NS BrDe; Mas (Mar; Sha; Cal); Cal (Mor); ArDi; FoAs 
Evans et al. (35), UK, 1978–1992 University hospital 126 NS (NS) 126 NS (NS)    IHC NOS (AP21N)  39 (31) NS Analog Yes Cal (Mor; Dis) 
Fasching et al. (36), Germany, 2001–2003 University hospital 423 60a (±13) 423 1.9a (1.4)    IHC NOS  37 (9) Mammomat 3000 Nova (Siemens) Analog Yes BrDe 
Gajdos et al. (37), USA, 1993–1999 Tertiary care hospital 118 59a (30–92) 118 1.0a (0.1–3.0)    IHC NOS NS 36 (31) NS Analog Yes Mas (Cal); Cal; ArDi 
Gu et al. (38), China, 2003–2005 University hospital 62 51a (31–79) 62 NS (NS)    IHC NOS  18 (29) Senographe DMR (GE) Analog NS NS Cal 
Jiang et al. (39), China, 2008–2010 University hospital 63 49a (27–79) 63 NS (0.5–9.0)    IHC Golden Bridge International (CB11)  19 (30) Mammomat Novation DR (Siemens) Digital Yes Mas (Mar
Kim et al. (40), South Korea, 2005–2008 Cancer center 151 48a (26–75) 151 NS (NS)    IHC Zymed (NS)  54 (36) Selenia (Hologic) Digital NS Cal 
Ko et al. (41), South Korea, 2007–2008 Cancer center; only TN, ER+/PR/HER2, ER/PR/HER2+ 239 52a (26–86) 239 2.3a (NS)    IHC Zymed (NS)  63 (26) Selenia (Hologic) Digital NS Mas (Cal); Cal; ArDi; FoAs 
Kuo et al. (42), Taiwan, 1999–2005 University hospital; <stage IV disease, follow-up ≥ 1 y 998 51a (±11) 998 NS (NS) NOS  IHC BioGenex (CB11); FISH NOS  156 (16) NS NS NS Level of suspicion 
Li and Chen (43), China, 2005–2009 University hospital 110 52a (29–77) 110 NS (NS)    IHC NOS  49 (45) Giotto NOS (IMS) Digital Yes Mas (Mar); Cal (Mor
Ma et al. (44), USA, 1994–1998 Women's Contraceptive and Reproductive Experiences Study 352 49a (35–64) 352 NS (NS) NOS  IHC Dako (10H8)  62 (18) NS Analog NS BrDe 
Månsson et al. (45), Sweden, 1986–1995 Population-based; unifocal, ≤1.5 cm 361 63b (±12) 361 NS (0.1–1.5) NOS  IHC Dako (polyclonal) Other 142 (39) NS Analog Yes Cal 
Mun et al. (46), South Korea, 2007–2010 University hospital; screen-detected 213 50a (30–76) 213 2.4a (0.5–8.5)    IHC NOS; FISH NOS  77 (36) Senographe DS; Senographe Essential (GE) Analog; digital NS Cal 
Palka et al. (47), Hungary, 2000–2003 University hospital; high-risk breast cancer adjuvant study 55 52a (33–70) 55 3.5a (NS) NOS  IHC NOS  14 (25) NS NS Yes Cal 
Palka et al. (48), Hungary, 2004–2006 University hospital 560 58a (28–85) 560 NS (NS)    IHC NOS  85 (15) NS NS ≥2 NS Cal 
Phipps et al. (49), USA, 1999–2008 BCSC; no history of breast cancer 7,281 NS (NS) 7,281 NS (NS)    IHC NOS; FISH NOS  ER/PR− 172 (2) NS NS NS Yes BrDe 
Pollan et al. (50), Spain, 1996–2005 Navarre Breast Cancer Screening Program 834 NS (NS) 834 NS (NS)    NS NS 111 (13) NS NS Yes BrDe 
Seo et al. (51), USA, 2001–2003 University hospital 498 57a (24–95) 543 NS (NS)    IHC Dako (HercepTest); FISH PathVysion  159 (29) Mammomat 3000 Nova (Siemens), Seno-graphe 2000D (GE) Analog, digital NS BrDe; Mas (Mar; Sha; Cal); Cal (Mor; Dis); ArDi; FoAs; SkTh; Oth 
Shin et al. (52), South Korea, 2005–2006 University hospital; node negative 710 49a (21–81) 715 NS (NS)    IHC NOS  165 (23) Senographe DMR (GE) Analog Yes Mas (Mar; Cal); Cal 
Sun et al. (53), China, 2004–2007 General hospital 144 49a (23–83) 144 NS (NS) NS IHC Beijing Zhongshan Biotechnology (NS) NS 58 (40) NS Digital NS Cal 
Taneja et al. (54), UK, 1996–1997 University hospital 415 55a (27–70) 415 NS (NS)    IHC Dako (NS) Other 132 (32) NS Analog Yes BrDe; Mas (Mar); Cal; ArDi; FoAs 
Wang et al. (55), China, 2003–2006 University hospital 152 51b (27–82) 152 NS (NS)    IHC Dako (K5204)  73 (48) Gold Standard (Pan-pacific Enterprises Inc.) Digital Yes Mas (Mar; Cal); Cal (Mor; Dis); ArDi; FoAs 
Wang et al. (56), USA, 1997–2003 University hospital; only ER 56 53a (31–84) 56 2.4a (0.2–8.0)    IHC Dako (HercepTest); FISH PathVysion  23 (41) NS Analog Yes BrDe; Mas (Mar; Sha; Cal); Cal (Mor); ArDi; FoAs 
Wang et al. (57), China, 2006–2007 General hospital 60 50b (28–75) 60 NS (NS)    IHC Fuzhou Maixin (NS)  30 (50) Giotto NOS (IMS) NS NS Yes Mas (Cal); Cal; ArDi 
Wen et al. (58), China, 2010 University hospital; no [ER+ or PR+ and HER2+] group 477 55b (25–86) 477 2.2b (0.5–7.0)    IHC NOS  76 (16) NS Digital ≥2 NS Mas (Mar; Sha; Cal); Cal (Mor; Dis); SkTh; Oth 
Yaghjyan et al. (59), USA, 1989–2004 Nurses' Health Study; postmenopausal 563 NS (NS) 563 NS (NS)    IHC NOS  140 (25) NS Analog NS Yes BrDe 
Yang et al. (60), USA, 1999–2005 Cancer center; ≤45 y premenopausal 198 34a (25–46) 198 2.8a (0.7–9.0)    IHC NOS; FISH NOS  67 (34) M3 (Hologic), Senographe DMR (GE) Analog Yes BrDe; Mas (Mar; Sha; Cal); Cal; ArDi 
Breast ultrasound             Transducer bandwidth    
Agrawal et al. (28), Poland, 1992–2001 University hospital 184 57a (32–75) 184 NS (NS)    IHC Dakopatts (K-353) NS 100 (54) NS NS NS NS Level of suspicion 
Au-Yong et al. (61), UK, 1996–1997 University hospital 227 63a (27–70) 227 NS (NS)    IHC Dako (NS) Other 71 (31) NOS (Esaote) 7.5 MHz Yes Mas (Sha; Ori; Mar; Hal; EcPa; PAF; Cal; Mul); Cal; Oth 
Chen et al. (14), China, 2006–2007 Military hospital 73 54a (33–85) 73 NS (0.5–6.6)    IHC Beijing Zhongshan Biotechnology (NS) NS 39 (53) Logiq7 (GE) 8.0–10.0 MHz NS NS Mas (Ori; Mar; Hal; PAF; Cal) 
Chen et al. (62), Taiwan, 2008–2009 Cancer center 168 51a (24–87) 168 2.6a (0.7–5.3)    IHC Dako (NS); FISH NOS  44 (26) Voluson 730 (GE) 6.0–12.0 MHz Computer-automated Vascularization 
Di Nublia et al. (33), Italy, 2000–2002 Cancer center; ≤ 35 y, no prior breast cancer 238 32a (22–35) 238 2.3a (NS)    IHC Triton (TAB250)  66 (28) Technos MPX, AU5 Harmonic (Esaote) 7.5–13.0 MHz NS NS Level of suspicion 
Kim et al. (63), South Korea, 2002–2006 University hospital 458 56a (25–87) 458 2.1a (0.4–12.5)    IHC Zymed (TAB250)  191 (42) Logiq9 (GE); HDI-5000 (Philips) 5.0–14.0 MHz 1∣2 Yes Mas (Sha; Ori; Mar; Hal; EcPa; PAF; Cal
Ko et al. (41), South Korea, 2007–2008 Cancer center; only TN, ER+/PR/HER2, ER/PR/HER2+ 245 52a (26–86) 245 2.3a (NS)    IHC Zymed (NS)  65 (27) HDI-5000, IU-22 (Philips) 5.0–12.0 MHz NS Mas (Sha; Ori; Mar; Hal; EcPa; PAF
Li et al. (15), China, 2007–2008 University hospital 120 47a (25–80) 120 NS (NS)    IHC MaxVision (NS)  99 (82) Logiq9 (GE); Acuson Sequoia 512 (Siemens) 7.0–14.0 MHz NS NS Mas (Mar; Hal; PAF; Cal); Vascularization 
Pang et al. (64), China, (2007–2008) University hospital 207 49b (25–79) 207 NS (NS)    IHC NOS; FISH NOS  62 (30) DU8 (Esaote); HDI-5000 (Philips) 4.0–13.0 MHz NS NS Mas (Mar; PAF; Cal); Vascularization 
Shin et al. (52), South Korea, 2005–2006 University hospital; node negative 710 49a (21–81) 715 NS (NS)    IHC NOS  165 (23) HDI-5000; IU-22 (Philips) 5.0–12.0 MHz Yes Mas (Sha; Cal); Cal 
Wan et al. (65), China, 2009–2011 University hospital 74 55a (32–78) 74 2.5a (1.1–5.3)    IHC Dako (A0485)  25 (34) DU8 (Esaote) 4.5–7.5 MHz Yes Enhancement patterns 
Wang et al. (56), USA, 1997–2003 University hospital; only ER 32 53a (31–84) 32 2.4a (0.2–8.0)    IHC Dako (HercepTest); FISH (PathVysion)  17 (53) NS NS Yes Mas (Sha; Ori; Mar; EcPa; Cal); Vascularization 
MRI             Acquisition and protocol    
Agrawal et al. (66), USA, 2003–2005 University hospital 78 NS (NS) 80 1.7a (±1.1)    NS NS 30 (38) 1.5T NOS (Philips) Bilateral; breast coil; DCE Yes Mas (MuFo; Sha; Mar; InEn); KiCu; TtP 
Baek et al. (16), USA, 2005–2006 University hospital; ≥1.5 cm only 66 51a (32–76) 66 3.4a (1.5–8.6)    IHC NOS; FISH NOS  21 (32) 1.5T Eclipse (Philips) Bilateral; breast coil; MRS NS NS Cho 
Baltzer et al. (67), Germany, 2005–2006 University hospital 128 60a (30–82) 145 NS (NS)    IHC Dako (HercepTest)  43 (30) 1.5T MAGNETOM Symphony (Siemens) Bilateral; breast coil; DCE NA NA Computer aided DCE pattern 
Chang et al. (68), South Korea, 2004–2006 University hospital 136 48a (26–75) 151 1.9a (0.5–5.8)    IHC NOS  39 (26) 1.5T MAGNETOM Sonata (Siemens) NS; breast coil; DCE NS Mas (Sha; Mar; InEn); KiCu; TtP 
Chen et al. (69), USA, 2006–2010 University hospital; neoadjuvant therapy 50 49a (28–82) 50 4.5a (1.7–11.8)    FISH NOS Other 17 (34) 3T Achieva (Philips) Bilateral; breast coil; DCE Yes Mas 
Choi et al. (70), South Korea, 2008–2010 University hospital 117 52a (29–81) 117 2.8a (±2.2)    IHC NOS  50 (43) 1.5T Achieva (Philips); 3T MAGNETOM Verio (Siemens) NS; breast coil; DWI NS ADC 
Costantini et al. (71), Italy, 2007–2011 University hospital 225 53a (±12) 225 3.4a (±2.0)    IHC Dako (polyclonal); FISH Dako (pharmDx)  38 (17) 1.5T Signa Excite (GE) Bilateral; breast coil; DCE, DWI NS Mas (MuFo; Sha; Mar; InEn); KiCu; T2Si; ADC; Oth 
Fernández-Guinea et al. (72), Spain, 1999–2006 General hospital 68 56a (30–83) 68 2.3a (0.6–12.0)    IHC Dako (polyclonal); FISH NOS  27 (40) 1.5T Signa EchoSpeed (GE) Bilateral; breast coil; DCE NS Mas (Mar; InEn); KiCu; TtP; EnRa 
Girardi et al. (73), Italy, 2006–2008 University hospital 72 54a (33–87) 72 2.2a (0.6–7.0)    IHC NOS  12 (17) 1.5T MAGNETOM Symphony (Siemens) Bilateral; breast coil; DCE NS Mas (Sha; Mar; InEn; KiCu; EnRa; Fis) 
Gómez-Raposo et al. (74), Spain, 2008–2011 University hospital; neoadjuvant therapy 24 NS (NS) 24 4.4 (2.4–6.4)f    IHC NOS NS 6 (25) 3T Achieva (Philips) NS; NS; DCE NS NS Mas; KiCu 
Jeh et al. (75), South Korea, 2008–2009 University hospital 107 52a (31–81) 107 3.2a (0.6–16.0)    IHC NOS  48 (45) 1.5T Achieva (Philips); 3T MAGNETOM Verio (Siemens) NS; breast coil; DCE NS Mas (Sha; Mar; InEn
Kim et al. (76), South Korea, 2005–2008 University hospital 62 47a (26–70) 62 NS (NS)    NS NS 30 (48) 1.5T MAGNETOM Avanto (Siemens) Bilateral; breast coil; DCE Yes TtP; Oth 
Kim et al. (77), South Korea, 2007 University hospital 62 54a (36–86) 62 2.2b (NS)    IHC Novocastra (NS)  15 (24) 1.5T Signa TwinSpeed (GE) NS; breast coil; DWI NS ADC 
Koo et al. (78), South Korea, 2009–2010 University hospital 68 54a (33–77) 68 1.9a (0.1–5.5)    IHC NOS; FISH NOS  10 (15) 1.5T Signa HDx (GE) Bilateral; breast coil; DCE NS Perfusion parameters 
Lee et al. (79), South Korea, 2004 University hospital 194 47a (26–76) 194 2.2a (0.1–6.0)    IHC Novocastra (NS)  24 (12) 1.5T MAGNETOM Sonata (Siemens) Unilateral; breast coil; DCE Yes Mas (Sha; Mar; InEn); KiCu; TtP; EnRa; Oth 
Liu and Peng (80), China, 2007–2009 University hospital 41 51a (36–70) 41 2.6a (NS)    IHC NOS  29 (71) 1.5T Signa TwinSpeed (GE); 3T Signa HDxt (GE) Bilateral; breast coil; DCE Yes Mas 
Loo et al. (81), the Netherlands, 2000–2008 Cancer center; neoadjuvant therapy 188 46a (23–76) 188 5.0a (1.4–11.2)    IHC NOS; CISH NOS  38 (20) 1.5T MAGNETOM Vision (Siemens); 3T Achieva (Philips) Bilateral; breast coil; DCE Yes Mas (MuFo) 
Lu et al. (82), China, 2005–2010 University hospital; mammography occult–axillary metastasis 17 53a (30–72) 17 1.5a (0.2–3.5)    IHC NOS Other 7 (41) 1.5T Signa Infinity Excite II (GE) NS; breast coil; DCE NS Mas (Mar; InEn); KiCu; Oth 
Makkat et al. (83), Belgium, 2004–2005 University hospital 57 50b (31–80) 57 2.2b (NS)    IHC Dako (HercepTest); FISH (PathVysion)  13 (23) 1.5T Intera (Philips) Bilateral; breast coil; DCE NA Tumor blood flow 
Marcos de Paz et al. (84), Spain, 2003–2009 University hospital 69 53a (31–74) 75 NS (NS)    IHC NOS; FISH NOS  29 (39) NS Signa (GE); NS Excite (GE) NS; NS; DCE NS NS Mas; KiCu; EnRa; Oth 
Martincich et al. (85), Italy, 2007–2009 Cancer center 190 47b (27–74) 192 2.2a (0.6–6.5)    IHC Dako (HercepTest); FISH NOS  40 (21) 1.5T NOS (GE) NS; breast coil; DWI NS ADC 
Montemurro et al. (86), Italy, 2003–2005 Cancer center 75 51b (27–74) 75 3.6b (0.7–12.0)    IHC Dako (HercepTest)  16 (21) 1.5T NOS (GE) NS; breast coil; DCE NS Sha; Mar; InEn; KiCu; EnRa; Fis 
Nakajo et al. (87), Japan, 2007–2008 General hospital 35 58a (35–77) 35 1.6a (0.5–3.0)    IHC Dako (HercepTest)  7 (20) 1.5T Intera Achieva (Philips) Unilateral; breast coil; DWI NS ADC 
Sah et al. (88), India, 2007–2011 University hospital 47 46a (25–75) 47 NS (NS)    IHC NOS  13 (28) 1.5T MAGNETOM Avanto (Siemens) NS; breast coil; MRS NS Cho 
Shin et al. (89), South Korea, 2008–2010 University hospital; ≥1.0 cm only 184 48a (28–72) 184 3.5a (1.0–10.0)    IHC NOS; FISH NOS  50 (27) 1.5T MAGNETOM Avanto (Siemens) NS; breast coil; MRS NS Cho 
Szabó et al. (90), Sweden, NS University hospital 61 47a (34–78) 60 2.4a (0.3–7.0)    IHC Dako (A0485)  14 (23) 1.5T MAGNETOM SP 63 (Siemens) Bilateral; breast coil; DCE Yes InEn; KiCu; TtP 
Tozaki and Hoshi (91), Japan, 2008–2009 General hospital; ≥1.5–3.0 cm only 50 52a (30–86) 50 2.2a (1.5–3.0)    IHC Dako (NS); FISH (PathVysion)  6 (12) 1.5T MAGNETOM Avanto (Siemens) Bilateral; breast coil; MRS NS NS Cho 
Youk et al. (92), South Korea, 2009–2011 University hospital 269 50a (26–83) 271 2.4a (0.5–10.0)    IHC Ventana (NOS); FISH (PathVysion)  94 (35) 1.5T MAGNETOM Avanto (Siemens); 3T Intera Achieva (Philips) Bilateral; breast coil; DCE, DWI Yes Mas (Sha; Mar; InEn); KiCu; T2Si; ADC; Oth 
FDG-PET             Acquisition and protocol    
Choi et al. (70), South Korea, 2008–2010 University hospital 117 52a (29–81) 117 2.8a (±2.2)    IHC NOS  50 (43) Biograph Duo or Truepoint (Siemens) FOV, NS; slice, 4.5–6.5 mm; res, NOS; fasting, NS; time, NS; dose, NS NS SUVmax 
Garcia Vicente et al. (93), Spain, NS Seven hospitals; neoadjuvant therapy 60 53a (±11.7) 60 4.3a (±2.1)    IHC NOS; FISH NOS  22 (37) Discovery DST-E, 16s (GE) FOV, NS; slice, 3.8 mm; res, NS; fasting, 4 h; time, 60′; dose, 370 MBq Yes SUVmax 
Gil-Rendo et al. (17), Spain, 2000–2004 University hospital 251 51a (24–87) 251 2.3b (0.3–4.9)    IHC Dako (A0485)  116 (46) ECAT EXACT HR+ (Siemens) FOV, 15.5-cm axial; slice, 3.9 mm; res, 4.5 mm; fasting, 6 h; time, 40–60′; dose, 370 MBq NS SUVmax 
Groheux et al. (94), France, 2006–2009 General hospital; only T2, T3 or T4 130 50b (26–81) 130 5.1b (2.1–14.0)    IHC Novocastra (CB11)  24 (18) Gemini XL (Philips) FOV, NS; slice, NS; res, NS; fasting, 6 h; time, 60′; dose, 5 MBq/kg Yes SUVmax 
Heudel et al. (18), France, 2008–2009 Cancer center 45 55b (26–85) 45 2.5b (0.8–9.0)    IHC Dako (A0485)  9 (20) Gemini (Philips) FOV, NS; slice, NS; res, NS; fasting, 6 h; time, 60′; dose, 5 MBq/kg NS NS SUVmax 
Humbert et al. (95), France, NS Cancer center; large or locally advanced 115 51a (NS) 115 NS (NS)    IHC Ventana (4B5); FISH (ZytoLight)  37 (32) C-PET Plus or Gemini GXL (Philips) FOV, NS; slice, NS; res, NS; fasting, 6 h; time, 60–90′; dose, 2–5 MBq/kg NS NS SUVmax 
Ikenaga et al. (96), Japan, 2004–2005 General hospital 47 63a (43–87) 47 2.1a (0–11.9)    IHC Dako (HercepTest)  14 (30) Discovery LS (GE) FOV, NS; slice, 4.25 mm; res, 5.4 mm; fasting, 6 h; time, 55–65′ after 3.7 MBq/kg Yes SUVmax 
Jin et al. (97), South Korea, 2006–2010 University hospital; neoadjuvant therapy no trastuzumab 273 44a (23–78) 273 4.9a (0.9–15.1)    IHC NOS  72 (26) Discovery STE (GE), Biograph 16 or Biograph Truepoint 40 (Siemens) FOV, NS; slice, NS; res, NS; fasting, NS; time, 60′; dose, 370–555 MBq No SUVmax 
Keam et al. (98), South Korea, 2006–2008 University hospital; neaoadjuvant phase II trial 78 45b (26–69) 78 4.5b (2.0–11.0)    IHC NOS; FISH NOS  17 (22) Gemini (Philips) FOV, NS; slice, NS; res, NS; fasting, 6 h; time, 50–75′; dose, 5.18 MBq/kg NS NS SUVmax 
Kim et al. (99), South Korea, 2009–2010 University hospital; ≥1.0 cm 91 49a (26–81) 91 2.2b (1.0–6.5)    IHC Dako (NS); FISH NOS  17 (19) Allegro (Philips) FOV, NS; slice, NS; res, NS; fasting, 6 h; time, 60′; dose, 5.18 MBq/kg NS NS SUVmax 
Koolen et al. (100), the Netherlands, 2008–2011 Cancer center; neoadjuvant, >3.0 cm and/or positive node 214 50b (41–57)i 214 NS (NS)    IHC NOS; CISH NOS  53 (25) Gemini TF (Philips) FOV, NS; slice, 2 mm; res, 2 mm × 2 mm; fasting, 6 h; time, 60′; dose, 180–240 MBq NS SUVmax 
Mavi et al. (101), USA, NS University hospital, >0.5 cm 116 51b (24–80) 116 NS (0.5-NS)    IHC Dako (HercepTest)  20 (17) Allegro (Philips) FOV, NS; slice, 4 mm; res, NS; fasting, 4 h; time, 60′; dose, 2.5–5.2 MBq/kg NS SUVmax 
Nakajo et al. (87), Japan, 2007–2008 General hospital; >5 mm 35 58a (35–77) 35 1.6a (0.5–3.0)    IHC Dako (HercepTest)  7 (20) Discovery STE (GE) FOV, NS; slice, NS; res, 5.1 mm; fasting, 5 h; time, 60′; dose, 3.7 MBq/kg NS SUVmax 
Osborne et al. (102), USA, 2001–2004 Cancer center; >5.0 cm and/or positive node 32 50a (28–72) 32 3.0a (0.2–8.4)    IHC NOS; FISH NOS  13 (41) Advance or Discovery LS (GE), Biograph (Siemens) FOV, NS; slice, NS; res, NS; fasting, 6 h; time, 50–60′; dose, 370–555 MBq NS NS SUVmax 
Sanli et al. (103), Turkey, 2009–2011 University hospital 79 53a (21–81) 79 3.0b (0.7–10.0)    IHC Thermo scientific (SP3)  28 (35) NS FOV, NS; slice, NS; res, NS; fasting, 6 h; time, 60′; dose, 370–550 MBq NS NS SUVmax 
Straver et al. (104), the Netherlands, 2007, NS Cancer center; >3.0 cm and/or positive node 37 49a (30–68) 37 NS (NS)    NS NS 11 (30) Gemini TF (Philips) FOV, NS; slice, NS; res, 2 mm; fasting, 6 h; time, 60′; dose, 180–240 MBq NS SUVmax 
Tozaki and Hoshi (91), Japan, 2008–2009 General hospital; ≥1.5–3.0 cm only 50 52a (30–86) 50 2.2a (1.5–3.0)    IHC Dako (NS); FISH (PathVysion)  6 (12) Discovery ST (GE) FOV, 15.7-cm axial; slice, 3.3 mm; res, NS; fasting, NS; time, NS; dose, 4.3 MBq/kg NS NS SUVmax 
Ueda et al. (105), Japan, 2005–2007 University hospital 145 54a (34–81) 145 NS (NS)    IHC Dako (NS)  31 (21) Biograph LSO Emotion (Siemens) FOV, NS; slice and res, NS; fasting, 4 h; time, 60′; dose, 3.7 MBq/kg NS NS SUVmax 
Ueda et al. (106), Japan, 2007–2008 University hospital; only ER+ neoadjuvant therapy 11 74a (±9.3) 12 2.6a (±1.0)    IHC NOS  2 (17) Biograph Duo (Siemens) FOV, 15.8 cm; slice, 2.5 mm; res, 6.3 mm; fasting, 6 h; time, 60′; dose, 3.7 MBq/kg NS SUVmax 
Wang et al. (107), USA, 2007–2009 University hospital 98 53a (±13) 100 3.5a (0.5–13.0)    IHC NOS; FISH NOS  18 (18) (Dedicated) PEM Flex Solo (Naviscan) FOV, 24 cm ×16.4 cm; slice, NS; res, NS; fasting, 6 h; time, 121′; dose, 596 MBq Yes SUVmax 
PatientsBreast cancersHER2 assessmentImaging acquisition and results
HistologycPositiveReaders
Ref., country, recruitment periodRecruitmentNAge in years (range/SD)NSize in cm (range/SD)IDCElseDCISAssaydCutoffN (%)Imaging system (manufacturer)Imaging parametersNBlindedImaging features described in reviewe,f,g,h
Mammography             Analog/Digital    
Agrawal et al. (28), Poland, 1992–2001 University hospital 184 57a (32–75) 184 NS (NS)    IHC Dakopatts (K-353) NS 100 (54) NS Analog NS NS Level of suspicion 
Aiello et al. (29), USA, 1988–1995 Breast cancer screening program 461 NS (NS) 461 NS (0.1–10) NOS  IHC NOS Other 79 (17) NS Analog NS BrDe 
Arora et al. (30), USA, 2005–2007 Cancer center; no prior breast cancer 1,323 56b (27–91) 1,323 1.5b (0.1–11)    IHC NOS NS 170 (13) NS Analog, digital NS NS BrDe 
Badra et al. (31), Greece, 1994–2004 University hospital; nonpalpable screen-detected 75 56b (35–74) 75 NS (NS)    IHC BioGenex (CB11)  34 (45) NS NS NS Cal 
Cui et al. (32), China, 2009–2011 University hospital; triple-negative and triple-positive only 161 51a (27–87) 161 2.4a (NS)    IHC NOS  44 (27) Senographe 800T (GE) Analog Yes BrDe; Mas (Mar; Sha; Cal); Cal 
Di Nubila et al. (33), Italy, 2000–2002 Cancer center; ≤ 35 y, no prior breast cancer 232 32a (22–35) 232 2.3a (NS)    IHC Triton (TAB250)  64 (28) Senographe 2000D; Senographe DMR (GE) Analog, digital NS NS Level of suspicion; BrDe; Cal 
Enache et al. (34), Romania, 2010–2011 General hospital; only ER 40 51a (25–88) 40 4.1a (0.6–10.7)    IHC Dako (polyclonal); CISH Invitrogen  11 (28) NS NS NS NS BrDe; Mas (Mar; Sha; Cal); Cal (Mor); ArDi; FoAs 
Evans et al. (35), UK, 1978–1992 University hospital 126 NS (NS) 126 NS (NS)    IHC NOS (AP21N)  39 (31) NS Analog Yes Cal (Mor; Dis) 
Fasching et al. (36), Germany, 2001–2003 University hospital 423 60a (±13) 423 1.9a (1.4)    IHC NOS  37 (9) Mammomat 3000 Nova (Siemens) Analog Yes BrDe 
Gajdos et al. (37), USA, 1993–1999 Tertiary care hospital 118 59a (30–92) 118 1.0a (0.1–3.0)    IHC NOS NS 36 (31) NS Analog Yes Mas (Cal); Cal; ArDi 
Gu et al. (38), China, 2003–2005 University hospital 62 51a (31–79) 62 NS (NS)    IHC NOS  18 (29) Senographe DMR (GE) Analog NS NS Cal 
Jiang et al. (39), China, 2008–2010 University hospital 63 49a (27–79) 63 NS (0.5–9.0)    IHC Golden Bridge International (CB11)  19 (30) Mammomat Novation DR (Siemens) Digital Yes Mas (Mar
Kim et al. (40), South Korea, 2005–2008 Cancer center 151 48a (26–75) 151 NS (NS)    IHC Zymed (NS)  54 (36) Selenia (Hologic) Digital NS Cal 
Ko et al. (41), South Korea, 2007–2008 Cancer center; only TN, ER+/PR/HER2, ER/PR/HER2+ 239 52a (26–86) 239 2.3a (NS)    IHC Zymed (NS)  63 (26) Selenia (Hologic) Digital NS Mas (Cal); Cal; ArDi; FoAs 
Kuo et al. (42), Taiwan, 1999–2005 University hospital; <stage IV disease, follow-up ≥ 1 y 998 51a (±11) 998 NS (NS) NOS  IHC BioGenex (CB11); FISH NOS  156 (16) NS NS NS Level of suspicion 
Li and Chen (43), China, 2005–2009 University hospital 110 52a (29–77) 110 NS (NS)    IHC NOS  49 (45) Giotto NOS (IMS) Digital Yes Mas (Mar); Cal (Mor
Ma et al. (44), USA, 1994–1998 Women's Contraceptive and Reproductive Experiences Study 352 49a (35–64) 352 NS (NS) NOS  IHC Dako (10H8)  62 (18) NS Analog NS BrDe 
Månsson et al. (45), Sweden, 1986–1995 Population-based; unifocal, ≤1.5 cm 361 63b (±12) 361 NS (0.1–1.5) NOS  IHC Dako (polyclonal) Other 142 (39) NS Analog Yes Cal 
Mun et al. (46), South Korea, 2007–2010 University hospital; screen-detected 213 50a (30–76) 213 2.4a (0.5–8.5)    IHC NOS; FISH NOS  77 (36) Senographe DS; Senographe Essential (GE) Analog; digital NS Cal 
Palka et al. (47), Hungary, 2000–2003 University hospital; high-risk breast cancer adjuvant study 55 52a (33–70) 55 3.5a (NS) NOS  IHC NOS  14 (25) NS NS Yes Cal 
Palka et al. (48), Hungary, 2004–2006 University hospital 560 58a (28–85) 560 NS (NS)    IHC NOS  85 (15) NS NS ≥2 NS Cal 
Phipps et al. (49), USA, 1999–2008 BCSC; no history of breast cancer 7,281 NS (NS) 7,281 NS (NS)    IHC NOS; FISH NOS  ER/PR− 172 (2) NS NS NS Yes BrDe 
Pollan et al. (50), Spain, 1996–2005 Navarre Breast Cancer Screening Program 834 NS (NS) 834 NS (NS)    NS NS 111 (13) NS NS Yes BrDe 
Seo et al. (51), USA, 2001–2003 University hospital 498 57a (24–95) 543 NS (NS)    IHC Dako (HercepTest); FISH PathVysion  159 (29) Mammomat 3000 Nova (Siemens), Seno-graphe 2000D (GE) Analog, digital NS BrDe; Mas (Mar; Sha; Cal); Cal (Mor; Dis); ArDi; FoAs; SkTh; Oth 
Shin et al. (52), South Korea, 2005–2006 University hospital; node negative 710 49a (21–81) 715 NS (NS)    IHC NOS  165 (23) Senographe DMR (GE) Analog Yes Mas (Mar; Cal); Cal 
Sun et al. (53), China, 2004–2007 General hospital 144 49a (23–83) 144 NS (NS) NS IHC Beijing Zhongshan Biotechnology (NS) NS 58 (40) NS Digital NS Cal 
Taneja et al. (54), UK, 1996–1997 University hospital 415 55a (27–70) 415 NS (NS)    IHC Dako (NS) Other 132 (32) NS Analog Yes BrDe; Mas (Mar); Cal; ArDi; FoAs 
Wang et al. (55), China, 2003–2006 University hospital 152 51b (27–82) 152 NS (NS)    IHC Dako (K5204)  73 (48) Gold Standard (Pan-pacific Enterprises Inc.) Digital Yes Mas (Mar; Cal); Cal (Mor; Dis); ArDi; FoAs 
Wang et al. (56), USA, 1997–2003 University hospital; only ER 56 53a (31–84) 56 2.4a (0.2–8.0)    IHC Dako (HercepTest); FISH PathVysion  23 (41) NS Analog Yes BrDe; Mas (Mar; Sha; Cal); Cal (Mor); ArDi; FoAs 
Wang et al. (57), China, 2006–2007 General hospital 60 50b (28–75) 60 NS (NS)    IHC Fuzhou Maixin (NS)  30 (50) Giotto NOS (IMS) NS NS Yes Mas (Cal); Cal; ArDi 
Wen et al. (58), China, 2010 University hospital; no [ER+ or PR+ and HER2+] group 477 55b (25–86) 477 2.2b (0.5–7.0)    IHC NOS  76 (16) NS Digital ≥2 NS Mas (Mar; Sha; Cal); Cal (Mor; Dis); SkTh; Oth 
Yaghjyan et al. (59), USA, 1989–2004 Nurses' Health Study; postmenopausal 563 NS (NS) 563 NS (NS)    IHC NOS  140 (25) NS Analog NS Yes BrDe 
Yang et al. (60), USA, 1999–2005 Cancer center; ≤45 y premenopausal 198 34a (25–46) 198 2.8a (0.7–9.0)    IHC NOS; FISH NOS  67 (34) M3 (Hologic), Senographe DMR (GE) Analog Yes BrDe; Mas (Mar; Sha; Cal); Cal; ArDi 
Breast ultrasound             Transducer bandwidth    
Agrawal et al. (28), Poland, 1992–2001 University hospital 184 57a (32–75) 184 NS (NS)    IHC Dakopatts (K-353) NS 100 (54) NS NS NS NS Level of suspicion 
Au-Yong et al. (61), UK, 1996–1997 University hospital 227 63