SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure Statement
  8. References

Screening ultrasound (US) can increase the detection of breast cancer. However, little is known about the clinicopathologic characteristics of breast cancers detected by screening US. A search of the database for patients with breast cancer yielded a dataset in 6837 women who underwent breast surgery at Seoul National University Hospital (Korea). Of 6837 women, 1047 were asymptomatic and had a non-palpable cancer. Two hundred fifty-four women with 256 cancers detected by US (US-detected cancer) and 793 women with 807 cancers detected by mammography (MG-detected cancer) were identified. The imaging, clinicopathologic, and molecular data were reviewed. Univariate and multivariate analyses were carried out. Women with US-detected cancer were younger and were more likely to undergo breast-conserving surgery and to have node-negative invasive cancer (< 0.0001). By multivariate analysis, the significant independent characteristics were tumor size, mammographic density, final assessment category according to the American College of Radiology Breast Imaging Reporting and Data System, progesterone receptor (PR), human epidermal growth factor receptor 2 (HER2), and molecular subtype. Compared to tumors that were >2 cm in size, tumors that were ≤1 cm in size were 2.2-fold more likely to be US-detected cancers (= 0.02). Compared to the luminal A subtype tumors (estrogen receptor [ER]+, PR+, HER2−), luminal B subtype tumors (ER+, PR+, HER2 + ) were less likely to be in the US-detected cancer group (P < 0.01). Women with dense breasts were more likely to have US-detected cancer (< 0.01) versus those with non-dense breasts. Screening US-detected cancers were less likely to be diagnosed as category 5 instead of category 4 (< 0.01). In conclusion, women with US-detected breast cancer are more likely to have small-sized invasive cancer and more likely associated with the luminal A subtype. (Cancer Sci 2011; 102: 1862–1867)

The critical rationale for breast cancer screening is that early detection and resultant treatment improve the disease outcome.(1,2) Early detection often enables breast-conserving surgery and less harmful treatment. A large multicenter trial by the American College of Radiology Imaging Network (ACRIN) Protocol 6666 has shown that the addition of a single screening ultrasound (US) to mammography increased the detection of small and node-negative cancers from 7.6 to 11.9 per 1000 women with elevated risk for breast cancer.(3,4) A randomized controlled trial on effectiveness of US screening in women aged 40–49 years is underway in Japan.(5) Ultrasound has been proposed as a supplemental screening test in women with dense breast tissues and in high-risk women who cannot undergo MRI for any reason.(1) Although mammography remains the gold standard in breast imaging, high mammographic density may make breast cancer more difficult to detect and thus can increase the risk of cancer development in women screened by mammography alone.(3,6–14)

New individualized therapy approaches use specific molecular signatures, biomarkers, and clinicopathological features of tumors and patients.(15–19) Adjuvant systemic therapy is planned based on patient age, tumor size, histological grade, lymph node metastasis, hormonal receptor status, and human epidermal growth factor receptor 2 (HER2) status.(20) Comprehensive molecular analysis has revealed that breast cancer is not a single disease, but that it shows heterogeneous morphology and expression of various molecular markers.(16,18,21–27) Currently, gene expression studies have resulted in the identification of five molecularly distinct subtypes of breast cancer.(18,22,23) To date, these subtypes are luminal A and B, which are hormone receptor-positive, the HER2 subtype, the basal-like subtype, and an unclassified group. Molecular markers specific for breast cancer include estrogen receptor (ER), progesterone receptor (PR), c-erbB2 (HER2), Ki-67, p53, bcl-2, basal cytokeratins, and E-cadherin.(18,21) Clinical differences between ER-negative and ER-positive cancers have been long recognized.(17,22–27) Most patients with ER-negative cancers do not benefit from antihormonal therapy, and many of these cancers are unaffected by conventional cytotoxic chemotherapy.(15,17) Triple-negative cancer, defined as a tumor that is negative for ER, PR, and HER2, has a relatively poor prognosis.(19,28,29)

Although several previous reports, including the ACRIN study, have shown that US screening yielded an 0.28–0.46% increase in cancer detection rate and that most detected cancers were stage I invasive cancers,(4,8,11,30–32) less is known about the characteristics of US-detected breast cancer. In particular, the frequency of the various molecular markers and phenotypes among breast cancers detected with screening US has not been widely studied. The purpose of the current study, therefore, was to compare the imaging, clinicopathologic, and molecular characteristics of 254 asymptomatic women who had mammographically occult breast cancer detected only by US to the characteristics of 793 women with cancer detected by screening mammography.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure Statement
  8. References

Study population.  A search of the computerized database for patients with histopathologic diagnosis of breast cancer between September 2003 and April 2010 yielded a dataset of 6837 women who underwent breast cancer surgery at Seoul National University Hospital (Korea). The clinical records, radiologic images, and reports were retrospectively reviewed by a radiologist with 6 years of experience in breast imaging (H.R.K) to evaluate which methods resulted in the detection of breast cancer among mammogram, US, and physical finding. The cancer was determined to be found by US screening if the case met the following three conditions: (i) negative mammogram report; (ii) according to the clinical record, the patient had no obvious symptoms and no remarkable findings at a clinical breast examination carried out by an experienced breast surgeon; and (iii) the lesion detected by US led to a biopsy, which led to a diagnosis of breast cancer. Of the 6837 women, a total of 264 women with 266 cancers were attributed to the US-detected cancer group (two patients had bilateral cancers). In our institution, US screening has been provided to women with dense breast tissues and who have familial or personal history of breast cancer.

All mammographic images from the US-detected group were reviewed by two breast radiologists with 6 and 24 years of experience (H.R.K. and W.K.M., respectively) to evaluate whether or not the negative mammography report had suspicious findings in retrospect. Cases in which the mammography findings were identified as a correlate to US-detected cancer were excluded upon the consensus of two breast radiologists. Ten patients showed mammographic abnormalities, such as asymmetry or architectural distortion. In addition, 793 women who had mammographically visible but clinically non-palpable breast cancers were allotted to the mammography (MG)-detected cancer group. Ultrasound images of these 793 women were available and 14 had bilateral cancers. In total, 254 women with 256 cases of breast cancer constituted the US-detected group, and 793 women with 807 cases of breast cancer constituted the MG-detected group.

Imaging.  An institutional review board-approved retrospective review of the records from 1047 mammography and US images acquired from September 2003 to April 2010 was carried out. Mammographic breast compositions were reviewed by two breast radiologists with 4 and 6 years of experience (M.S.B. and H.R.K., respectively). Breast compositions were classified on a four-point scale (4 =  extremely dense, >75%; 3 =  heterogeneously dense, 51–75%; 2 = scattered fibroglandular tissue, 25–50%; 1 = almost entirely fatty, <25%) according to the American College of Radiology Breast Imaging Reporting and Data System (ACR BI-RADS) breast density grade.(33) During the study period, bilateral whole breast US were carried out by several radiologists with specialty training in breast imaging (4–24 years of experience) using the following high-resolution US equipment: an EUB-8500 scanner (Hitachi Medical, Tokyo, Japan); an HDI 5000 scanner (Advanced Technology Laboratories, Bothell, WA, USA); a LOGIQ 700 scanner (GE Medical Systems, Milwaukee, WI, USA); or a Voluson 530D scanner (GE-Kretz, Zipf, Austria) equipped with a 14–6 MHz linear array transducer. Individual radiologists classified the lesions detected at US on a scale of 1–5 adapted from the BI-RADS final assessment category:(34) category 1, negative finding; category 2, benign finding; category 3, probably benign finding; category 4, suspicious finding; and category 5, highly suggestive of malignancy. Subcategories of category 4 were defined according to the probability of malignancy as follows: 3–10% in 4a (low suspicion); 11–50% in 4b (intermediate suspicion); and 51–94% in 4c (moderate suspicion). In this study, the BI-RADS final assessment categories were determined by the radiology reports.

Clinical, pathological, and immunohistochemical data.  The electronic medical records of 1047 women were reviewed to determine their age and menopausal status. The family histories and personal histories of breast cancer were determined from the medical records and radiology reports. A history of prior benign biopsies was also recorded. The surgical pathology records, including biopsy results, were reviewed to determine the tumor histology, tumor size, multifocality/multicentricity, histologic grade, nuclear grade, and axillary lymph node status. The type of surgery was determined from the breast surgery report. According to the sixth edition of the American Joint Committee on Cancer Manual(35), the pathologic tumor size for TNM classification includes only the measurement of the invasive component. Thus, we did not incorporate measurements of the non-invasive components into the tumor size. The Elston–Ellis grading system(36) was used for histologic tumor grading in which a score of 1–3 was assigned for tubule formation, pleomorphism, and mitotic count. The total score could range from 3 to 9, with a total of 3–5 representative of grade 1 (low grade), a total of six or seven representative of grade 2 (intermediate grade), and a total of eight or nine representative of grade 3 (high grade).

Immunochemical staining for ER, PR, HER2, p53, Ki-67, bcl-2, epidermal growth factor receptor (EGFR), and cytokeratin 5/6 (CK5/6) was carried out on paraffin sections cut from tissue microarray blocks in our institution during the study period. We could not obtain the results for the CK5/6 stainings from 2003 to 2005 because immunochemical staining for CK5/6 was not routinely carried out during this time period. Immunostained tissue microarray slides were evaluated for expression and overexpression of each molecular marker. Tumor cells that showed nuclear staining for ER or PR were considered ER-positive or PR-positive (>10% of positive tumor cell nuclei), respectively. Tumor cells were considered positive for HER2 protein overexpression when more than 10% of the cells showed moderate or strong membrane staining (2 + and 3 +). Cases were considered CK-positive or EGFR-positive if any cytoplasmic and/or membranous staining was found in the tumor cells. The Ki-67 labeling index (KI) was calculated as the percentage of positive tumor nuclei divided by the total number of tumor cells examined. For the statistical analysis, KI values of 20% or more were classified as high, and KI values lower than 20% were classified as low. Molecular marker data, including clinicopathologic and imaging data, were collected and recorded in spreadsheets (Access and Excel; Microsoft, Redmond, WA, USA).

Classification of molecular phenotype.  All tumors were classified into five molecular subtypes following the classification method proposed by Carey et al.(37) Cases that were ER-positive and/or PR-positive and HER2-negative were classified as luminal A type; cases that were ER-positive and/or PR-positive and HER2-positive were classified as luminal B type; cases that were ER-negative, PR-negative, and HER2-positive were classified as HER2 type; and cases that were negative for ER, PR, HER2, and positive for CK5/6 and/or EGFR were classified as basal-like. Cases that lacked expression of all five markers were considered unclassified.(38)

Statistical analysis.  Frequency distributions and simple means were generated to describe the study data. Categorical variables between groups were analyzed by the chi-square test or Fisher’s exact method where appropriate. Student’s t-test was used for continuous variables, such as age and tumor size. Multivariate logistic regression analysis was carried out to estimate the odds ratio and 95% confidence intervals of US-detected cancers in relation to imaging and clinicopathologic characteristics. Initially, univariate logistic analysis was used to identify candidate variables for the model and variables with P-values <0.2 were selected for the final model. All statistical analyses were carried out using spss version 10.0 for Windows (SPSS, Chicago, IL, USA) and P-values <0.05 were considered significant.

Results

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure Statement
  8. References

Univariate analysis.  Two hundred fifty-four women who underwent US screening had clinically and mammographically occult breast cancer. These women had a mean age of 48 years. Patients who had cancer detected by US were younger than patients who had cancer detected by mammography (mean age 48 vs 52 years, < 0.0001) (Table 1). Of the 254 women in the group with US-detected cancer, 152 (59.8%) were premenopausal, versus 311 (39.2%) of 793 women in the other group (< 0.0001). Women with US-detected cancer were more likely to report a family history of breast cancer (3.5 vs 2%; < 0.0001), personal history of breast cancer (2.8 vs 1.3%; < 0.0001), and previous benign breast disease (5.5 vs 1.8%; < 0.0001). One of 14 women in the US-detected group with previous benign biopsies had a high-risk lesion, such as atypical ductal epithelial hyperplasia. As expected, women with US-detected cancer were more likely to have dense breast tissues with heterogeneously or extremely dense breast compositions on mammography compared to women with MG-detected cancer (89.4 vs 64.9%; < 0.0001). In terms of the final BI-RADS assessment category, US-detected cancers were more likely to be diagnosed as category 4 compared to MG-detected cancers (73.6 vs 39.5%; < 0.0001).

Table 1.   Clinical and imaging characteristics of 254 women who had screening ultrasound (US)-detected breast cancer and 793 women who had mammography (MG)-detected breast cancer
CharacteristicsUS-detected, n (%)MG-detected, n (%)P-value
  1. †This group includes one case of atypical ductal hyperplasia. ‡US-detected group (n, %): grade 1 (2, 0.8%); grade 2 (25, 9.8%); grade 3 (143, 56.3%); grade 4 (84, 33.1%). MG-detected group (n, %): grade 1 (79, 10%); grade 2 (199, 25.1%); grade 3 (428, 54%); grade 4 (87, 10.9%). §Chi-square-test for category 4 and category 5 was also carried out (< 0.0001).

Age (years)
 Mean ± SD48 ± 752 ± 9<0.0001
 30–3923 (9.1)62 (7.8)
 40–49122 (48)267 (33.7)
 50–59 90 (35.4)295 (37.2)
 ≥6019 (7.5)169 (21.3)
Menopausal status
 Premenopausal152 (59.8)311 (39.2)<0.0001
 Peri- or postmenopausal102 (40.2)482 (60.8)
Family history of breast cancer 9 (3.5)16 (2)<0.0001
Personal history of breast cancer 7 (2.8)10 (1.3)<0.0001
Prior benign biopsy†14 (5.5)14 (1.8)<0.0001
Breast density (BI-RADS)‡
 Non-dense (grade 1 and 2) 27 (10.6)278 (35.1)<0.0001
 Dense (grade 3 and 4)227 (89.4)515 (64.9)
Final BI-RADS category
 3 2 (0.8)0 (0)<0.0001§
 4187 (73.6)313 (39.5)
  4a 79 (31.1)28 (3.5)
  4b 68 (26.8)142 (17.9)
  4c 40 (15.7)143 (18)
 5 65 (25.6)480 (60.5)

Women with US-detected cancer were more likely to undergo breast-conserving surgery (84.4 vs 68.4%; < 0.0001) and have a node-negative invasive cancer (91.8 vs 84.1%; < 0.0001) (Table 2). The mean tumor size was 1.32 cm in the US-detected group versus 1.70 cm in the MG-detected group (< 0.0001). Notably, the frequency of tumors with sizes of 1 cm or less was higher in the US-detected group (41.5 vs 32.2%; = 0.008). Multifocality or multicentricity in the surgical specimen was more common in MG-detected cancers than in US-detected cancers (15 vs 3.5%; < 0.0001). The tumor grading results in women with ductal carcinoma in situ (DCIS) and invasive cancer were found to be similar. For invasive breast cancer, US-detected cancers were more likely than MG-detected cancers to be well- or moderately-differentiated (82.6 vs 70.7%; = 0.0018) and were less likely to have a high nuclear grade (76.5 vs 65.7%; = 0.0177). The DCIS cases detected by US were also more likely to be low-grade or intermediate-grade lesions (54.2 vs 34.4%; = 0.0340).

Table 2.   Histopathologic characteristics of breast cancers detected by screening ultrasound (US-detected, = 256) and mammography (MG-detected, n = 807)
CharacteristicsUS-detected, n (%)MG-detected, n (%)P-value
  1. †Chi-square-test was carried out for tumors ≤1.0 cm and >1.0 cm (= 0.008). ‡Value of histologic grade in only invasive cancer is depicted in this table. Grade I, well differentiated (US-detected, n = 44, 22.6%; MG-detected, n = 94, 17.4%). Grade II, moderately differentiated (US-detected, n = 117, 60.0%; MG-detected n = 288, 53.3%). Grade III, poorly differentiated. Unknown, “not described in the pathology report”. §Grade 1, low grade (US-detected, n = 19, 9.3%; MG-detected, n = 44, 7.5%). Grade 2, intermediate grade (US-detected, n = 137, 67.2%; MG-detected, n = 341, 58.2%). Grade 3, high grade. Unknown, “not described in the pathology report”.

Type of surgery
 Breast-conserving surgery216 (84.4)552 (68.4)<0.0001
 Mastectomy 40 (15.6)255 (31.6)
Tumor histology
 Ductal carcinoma in situ 48 (18.8)214 (26.5)<0.0001
 Invasive carcinoma208 (81.2)593 (73.5)
Multifocality
 Monofocal247 (96.5)686 (85.0)<0.0001
 Multifocal 9 (3.5)121 (15.0)
Invasive tumor size (cm)
 Mean ± SD     1.32 ± 0.671.70 ± 1.05<0.0001†
 ≤1.0 85 (41.5)191 (32.2)
 1.1–2.0100 (48.3)265 (44.7)
 2.1–5.0 21 (10.1)130 (21.9)
 5.1-10.00 (0) 7 (1.2)
Histologic grade‡
 I or II161 (82.6)382 (70.7)0.0018
 III 34 (17.4)158 (29.3)
 Unknown1353
Nuclear grade
DCIS
 Low or intermediate 26 (54.2) 72 (34.4)0.0340
 High 22 (45.8)137 (65.6)
 Unknown05
Invasive carcinoma§
 1 or 2156 (76.5)385 (65.7)0.0177
 3 48 (23.5)201 (34.3)
 Unknown47
Lymph node metastasis
 None235 (91.8)679 (84.1)<0.0001
 1–317 (6.6) 96 (11.9)
 4–9 3 (1.2)24 (3)
 ≥10 1 (0.4)8 (1)

Compared with MG-detected cancers, US-detected cancers were more likely to be ER- positive (= 0.015) and PR-positive (< 0.0001) (Table 3). However, HER2-positivity was significantly more frequent among the MG-detected cancers than among the US-detected cancers (37.4 vs 25%; < 0.0001). The US-detected cancers were more likely to be positive for bcl-2 expression than MG-detected cancers (83.8 vs 77.5%; = 0.033). However, no statistically significant differences were found in expression of Ki-67 (= 0.423), p53 (= 0.117), or EGFR (= 0.090).

Table 3.   Molecular marker data from breast cancers detected by screening ultrasound (US-detected) and mammography (MG-detected)
CharacteristicsUS-detected, n (%)MG-detected, n (%)P-value
  1. ER, estrogen receptor; EGFR, epidermal growth factor receptor; HER2, human epidermal growth factor receptor 2; PR, progesterone receptor.

ER
 Negative47 (18.7)204 (25.7)0.015
 Positive204 (81.3)582 (74.3)
 Unknown521
PR
 Negative67 (26.3)342 (43.5)<0.0001
 Positive188 (73.7)444 (56.5)
 Unknown121
HER2
 Negative191 (75)497 (62.6)<0.0001
 Positive64 (25)297 (37.4)
 Unknown113
Ki-67
 Negative247 (97.6)753 (96.3)0.423
 Positive6 (2.4)29 (3.7)
 Unknown325
p53
 Negative68 (27.3)257 (32.9)0.117
 Positive181 (72.7)525 (67.1)
 Unknown725
Bcl-2
 Negative41 (16.2)176 (22.5)0.033
 Positive212 (83.8)605 (77.5)
 Unknown326
EGFR
 Negative248 (98.0)744 (95.6)0.090
 Positive5 (2.0)34 (4.4)
 Unknown329

Based on the immunostaining data from these five markers, the US-detected cancer was further classified as follows: 167 tumors (66.5%) were classified as luminal A; 52 (20.9%) were luminal B; 12 (4.8%) were HER2; 6 (2.4%) were basal-like; and 14 (5.6%) tumors were unclassifiable (Table 4). Five cases could not be classified (“unknown”) because of staining that could not be evaluated. Screening US-detected cancers were significantly more likely to be the luminal A type compared to MG-detected cancers (66.5 vs 49.6%; < 0.0001). In contrast, the HER2 phenotype was more frequently detected among MG-detected cancers than among US-detected cancers (10.2 vs 4.8%; = 0.009). The luminal B phenotype was more frequent among MG-detected cancers (26.6%) than among the US-detected cancers (20.9%), although this difference was not statistically significant (= 0.067). The unclassified phenotype was more frequent among MG-detected cancers than among US-detected cancers (12.6 vs 5.6%; = 0.002).

Table 4.   Frequency of molecular phenotypes among breast cancers detected by screening ultrasound (US-detected) and mammography (MG-detected)
ImmunophenotypeUS-detected, n (%)MG-detected, n (%)P-value
  1. Staining could not be evaluated. HER2, human epidermal growth factor receptor 2.

Luminal A167 (66.5)393 (49.6)<0.0001
Luminal B52 (20.9)211 (26.6)0.067
HER212 (4.8)81 (10.2)0.009
Basal-like6 (2.4)8 (1.0)0.097
Unclassified14 (5.6)100 (12.6)0.002
Unknown†514 

Multivariate model.  A multivariate logistic regression analysis was carried out to assess how individual characteristics relate to each other and to identify the independent characteristics that show statistically significant differences between US-detected and MG-detected cancers. Table 5 lists the adjusted odds ratios and 95% confidence intervals for the association between US-detected breast cancer and selected characteristics such as mammographic density, lymph node status, BI-RADS category, ER, PR, HER2, immunophenotypes, tumor size, and tumor grade. The final assessment of BI-RADS category, mammographic density, PR, HER2, molecular phenotype, and tumor size were significant independent variables that differentiated between US-detected and MG-detected cancers. Specifically, compared to the luminal A subtype after adjusting for the above-mentioned variables, US-detected cancers were less likely to have the luminal B subtype (< 0.01). Compared to tumors sized >2 cm, tumors sized ≤1 cm were 2.21 times more likely to be associated with US-detected cancer (= 0.02). Furthermore, tumors sized >1 cm to ≤2 cm were 2.25-fold more likely to be associated with US-detected cancer (= 0.01). Patients with dense breasts, that is, with mammographic density of 50% or more, were 3.71 times more likely to have a breast cancer detected by US alone (< 0.01) compared to those with non-dense breasts, that is, with mammographic density less than 50%. Compared to category 4, US-detected cancers were less likely than MG-detected cancers to be diagnosed as category 5 (< 0.01). Tumor grade and lymph node metastasis were not significant independent characteristics of US-detected breast cancer, although DCIS grade showed a trend towards statistical significance (= 0.08).

Table 5.   Multivariate logistic regression analysis for characteristics of breast cancers detected by screening ultrasound (US-detected)
CharacteristicsCrude OR (95% CI)Adjusted OR (95% CI)P-value
  1. †Model includes age, menopausal status, mammographic (mammo) density, lymph node (LN) status, American College of Radiology Breast Imaging Reporting and Data System (BI-RADS), molecular markers (estrogen receptor [ER], progesterone receptor [PR], and human epidermal growth factor receptor 2 [HER2]), and immunophenotypes. ‡Model includes age, menopausal status, mammo density, LN status, BI-RADS, molecular markers (ER, PR, and HER2), immunophenotypes, and ductal carcinoma in situ (DCIS) grade. §Model includes age, menopausal status, mammo density, LN status, BI-RADS, molecular markers (ER, PR, and HER2), immunophenotypes, tumor size, invasive ductal carcinoma (IDC) histologic grade, and nuclear grade. CI, confidence interval; OR, odds ratio; Ref, referent group.

All cancer†
BI-RADS category
 4RefRef 
 50.24 (0.17–0.32)0.24 (0.17–0.33)<0.01
Mammo density
 Non-denseRefRef 
 Dense4.51 (2.95–6.91)3.71 (2.33–5.91)<0.01
LN metastasis
 NoRefRef 
 Yes0.48 (0.30–0.79)0.64 (0.37–1.09)0.10
ER
 NegativeRefRef 
 Positive1.69 (1.17–2.45)1.02 (0.47–2.22)0.97
PR
 NegativeRefRef 
 Positive2.35 (1.71–3.24)2.00 (1.29–3.09)<0.01
HER2
 NegativeRefRef 
 Positive0.57 (0.41–0.78)0.55 (0.38–0.78)<0.01
Immunophenotype
 Non-luminalRefRef 
 Luminal1.98 (1.31–2.99)1.19 (0.47–3.02)0.72
 Luminal ARefRef 
 Luminal B0.57 (0.40–0.82)0.52 (0.35–0.77)<0.01
 HER20.35 (0.19–0.67)0.60 (0.19–1.96)0.40
 Basal-like1.77 (0.60–5.17)2.06 (0.40–9.96)0.37
 Unclassified0.37 (0.20–0.69)0.54 (0.16–1.81)0.32
DCIS only‡
DCIS grade
 Low or intermediateRefRef 
 High0.48 (0.25–0.92)0.47 (0.20–1.08)0.08
Invasive only§
Tumor size (cm)
 >2RefRef 
 1–22.52 (1.45–4.38)2.25 (1.19–4.24)0.01
 ≤13.66 (2.08–6.45)2.21 (1.13–4.31)0.02
Histologic grade
 IRefRef 
 II0.90 (0.59–1.37)1.10 (0.62–1.98)0.74
 III0.47 (0.28–0.79)0.81 (0.32–2.04)0.65
Nuclear grade
 1RefRef 
 20.87 (0.48–1.57)0.63 (0.28–1.40)0.25
 30.52 (0.28–0.99)0.54 (0.19–1.56)0.26

Discussion

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure Statement
  8. References

Screening mammography has limitations in women with dense breasts. The sensitivity of mammographic detection is reduced in women with dense breast tissues because breast cancers have a radiographic attenuation that is similar to that of fibroglandular tissues of the breast, and the presence of a non-calcified tumor might be undetectable if the mass is within a fibroglandular area. However, screening US could enable the detection of mammographically occult cancers in women with dense breasts because most breast cancers are relatively hypoechoic and can be well-visualized against a background of hyperechoic fibroglandular tissue.(9) Approximately 90% of the women with US-detected breast cancer had dense breast tissues with mammographic density of 50% or more. The vast majority (99.2%) of the screened women who had a tumor detected by US had sonographic BI-RADS category 4 or 5 lesions that warranted biopsy. The US-detected cancers were more likely to be categorized as suspicious findings (category 4, 73.6%), whereas cancers visible on mammography were more frequently classified as lesions with a high probability of malignancy (category 5, 60.5%) because those cancers may frequently be associated with suspicious mammographic calcifications, such as linear-branching and pleomorphic calcifications.

Consistent with previous studies, our results indicate that the majority of US-detected cancers are invasive and node-negative. The mean histologic size of the invasive cancer in our group was 1.32 cm. Early detection of non-palpable, mammographically-occult, and node-negative invasive cancers by US screening would reduce breast cancer mortality because invasive cancers not visible by mammography could be expected to present later on interval cancers with worse prognosis.(2,4) Several published reports(4,8,11,30–32) have reported that US screening yielded an increased detection rate of 2.8–4.6 cancers per 1000 women with dense breasts and negative mammograms. Recently, the ACR and the Society of Breast Imaging (SBI) have published guidelines for breast cancer screening by imaging.(7) The ACR and SBI recommend that US may have a role as a screening tool for high-risk women who have contraindications to MRI or for those whose levels of risk do not reach the level recommended for breast MRI screening by the ACR. However, the use of US screening in high-risk women adds no benefit over screening with mammography and MRI. Screening MRI is costly and leads to short-term follow-ups and many benign biopsies due to a high incidence of false-positive image interpretation. The disadvantages associated with screening US are low positive predictive value for biopsy recommendations, poor reproducibility, and operator dependency in carrying out the examination. We think that US screening may benefit certain women, those with dense breasts who are at high risk for breast cancer and cannot undergo MRI.

A strength of our study is the inclusion of clinicopathological features of the US-detected breast cancer in the screened population. Screening by breast US was effective in detecting cancer in 3.5% (9/254) of women with a family history of breast cancer, 2.8% (7/254) of women with a history of prior breast cancer, and 5.5% (14/254) of women who had benign biopsies. In addition, DCIS accounted for 18.8% of the cancers found. Our results may be superior to those from other studies because previously reported results have indicated a 9–14% frequency of DCIS in the cancers detected by US screening.(9,32,39,40) Izumori et al.(41) examined US findings and histological features of 47 DCIS cases diagnosed by US alone compared with 103 DCIS cases diagnosed by mammography or clinically. Most DCIS cases detected by US alone were localized, showed few extensive intraductal components, and were of low grade according to the Van Nuys classification. In the present study, DCIS cases detected by screening US were more likely to be of non-high nuclear grade than DCIS detected by screening mammography (= 0.034) even though DCIS grade was not a significant independent variable of the US-detected cancer by multivariate analysis (= 0.08).

To our knowledge, we present the first study to determine the status of various molecular markers and molecular phenotypes in breast cancer detected by US alone. We have shown that the prevalence of the molecularly defined phenotypes differed significantly between US-detected and MG-detected cancers. Screening US-detected cancers were more likely to have the luminal A phenotype than MG-detected cancers, whereas MG-detected cancers were more likely to have the HER2 phenotype. However, in a multivariate model after adjusting for independent clinicopathologic variables, the only significant independent variable to emerge was the luminal B subtype. Compared to the luminal A subtype, US-detected cancers were less likely than MG-detected cancers to be classified as the luminal B subtype. These data provide evidence that US-detected and MG-detected cancers are both molecularly heterogeneous. Although ER expression characterizes luminal-type breast cancers, luminal B cancers have a higher rate of tumor cell proliferation and poorer prognosis than luminal A cancers.(20) Dawood et al.(42) classified 1945 invasive breast cancers into one of five molecular phenotypes based on immunohistochemical assays and found that patients with HER2-type and basal-like cancers had worse survival relative to those with luminal A-type cancers. Therefore, our results suggest that patients with US-detected cancers may have better prognoses than those with MG-detected cancers.

In conclusion, we present a comprehensive series of analyses of the imaging, clinicopathological, and immunohistochemical characteristics of breast cancers detected by screening US. Most tumors discovered by US screening were smaller, invasive cancers with lower levels of suspicion of malignancy at US. These cancers were more likely associated with the luminal A subtype compared to luminal B subtype. Screening US-detected cancer can therefore benefit from early detection, which would allow breast conservation and the use of less toxic therapy. Although breast US screening is not widely accepted at present, screening with mammography and US may enable earlier detection of invasive cancer in certain women with denser breast tissues. Automated whole-breast US is becoming available and would be an ideal screening tool. We realize that more encouraging results indicating the efficacy and accuracy of such technology would be necessary to determine the effectiveness of US screening.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure Statement
  8. References

This study was supported by the Innovative Research Institute for Cell Therapy (A062260) and by the Korea Communications Commission, Korea, under the R&D program supervised by the Korea Communications Agency (KCA-2011-11911-01108).

References

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure Statement
  8. References