Distribution of molecular subtypes of breast cancer in Chinese women
Our study showed that the prevalence of the triple-negative subtype of breast cancer among Chinese women (12.9%) is similar to that in European populations (10-16%), but lower than in African-American population (20-21%). The HER2 subtype accounted for 13.7% of Chinese breast cancer cases, which is higher than the reported positivity (4-8%) in either European or African-American populations [
4,
6‐
8,
22,
23]. In our study, approximately 8% of breast cancer patients had weak positive or borderline staining (2+) for HER2, which was interpreted as an equivocal category that would be recommended for verification with fluorescent in situ hybridization (FISH) for therapeutic indication of trastuzumab (Herceptin) treatment [
15]. The FISH-derived amplification rate for the HER2 equivocal group (i.e., IHC 2+) has been observed to be approximately 25% in Western women [
24,
25]. If a similar rate is true for Chinese women, most HER2 equivocal cases would fall into either luminal A or triple-negative subtypes. Therefore, the frequencies of luminal A (48.6%) and triple-negative subtypes in our study could be underestimated. We compared ER and PR status for the HER2 borderline group with the HER2+ and HER2- groups, and found that ER+ and PR+ rates (67.4% and 57.7%, respectively) for the HER2 borderline group were more similar to that of the HER2- group (69.7% and 67.1%, respectively) than to the HER+ group (46.1% and 41.3, respectively), suggesting that the vast majority of cases in our HER2 borderline group should likely be classified in the HER2- group. Regardless of the true HER2+ rate for the borderline group, the prevalence of the HER2 subtype in this Chinese population is higher than in Western populations.
The prevalence of breast cancer subtypes appears to differ among different races or ethnicities. It has been well documented that the triple-negative subtype is most common among young African-American patients, while luminal A is most common among postmenopausal white women [
4,
6‐
9,
22,
23]. The increased risk for the triple-negative subtype in African-American women may due to parity and younger age at first full-term pregnancy, multiple live births without breastfeeding, use of medications to suppress lactation [
7], and intrinsic genetic variables, such as higher p53 expression [
6] and particularly high prevalence of founder mutations in BRCA1 or BRCA2 gene in young (<35 years) African-American women [
26]. In our study population, women with triple-negative breast cancer more frequently reported a family history of breast cancer than did women with other subtypes. This suggests that genetic factors may play a more important role in this molecular subtype of breast cancer. Since BRCA mutations in Chinese women are uncommon (1.1% each for BRCA1 and BRCA2) [
27]; other genetic contributors to the triple-negative subtype in Chinese women need to be investigated.
We found that HER2+ breast cancers account for 30% of all breast cancer cases in our study population, similar to a previous report from Shanghai (31%) and higher than the reports from Tianjin (26%) [
13,
14], Taiwan (21%) [
28], and the US (26%) [
9]. Consistent with our findings, one large, registry-based population study [
8] showed that HER2+ tumors are more common among Asian/Pacific Islanders (28%) than among non-Hispanic Whites (21%) or non-Hispanic Blacks (24%), but similar to Hispanics (26%) (Table
4). Another large population study further revealed that among Asian-Americans, Korean and Philipino women had the highest prevalence of HER2+ tumors (36% and 31%, respectively), followed by Vietnamese (29%) and Chinese (26%) women, while Japanese and South Asian women showed a prevalence of HER2+ tumors similar to non-Hispanic Whites and non-Hispanic Blacks (19-23%) [
9]. It was not clear why the prevalence of the HER2 subtype or of HER2+ tumors is higher among Chinese or Asian women compared with women of European ancestry or African-American women. Although it has been suggested that environmental factors might play an important role in the etiology of HER2+ breast cancers, variations in criteria used to determine HER2 status may also contribute the differences.
Table 4
Distribution of breast cancer subtypes in different ethnicities and in different geographical areas of China, %
| AA | 47.4 | 12.8 | 8.2 | 20.9 | 26.5 | 5.1 | 196 |
| Caucasian | 54.0 | 17.3 | 5.7 | 23.0 | 16.0 | 7.0 | 300 |
| Polish | 68.0 | 14.0 | 5.0 | 19.0 | 13.0 | | 1985 |
| Italy | 68.7 | 6.0 | 7.6 | 13.6 | 11.8 | 6.0 | 804 |
| Non-Hispanic White | 67.1 | 15.1 | 6.2 | 21.3 | 11.6 | | 39051 |
| Non-Hispanic Black | 48.9 | 14.2 | 9.8 | 24.0 | 27.0 | | 2936 |
| Hispanic | 56.2 | 16.5 | 9.7 | 26.2 | 17.6 | | 7673 |
| Asian/Pacific Islander | 59.3 | 18.5 | 9.9 | 28.4 | 12.3 | | 5215 |
| Other | 65.1 | 14.3 | 5.4 | 19.7 | 15.2 | | 315 |
| Non-Hispanic White | 69.6 | | | 18.7 | 11.7 | | 60498 |
| Non-Hispanic Black | 51.1 | | | 22.7 | 26.2 | | 5292 |
| Hispanic | 58.1 | | | 24.9 | 17.0 | | 14106 |
| Japanese | 69.9 | | | 19.5 | 10.7 | | 1136 |
| Chinese | 63.5 | | | 25.6 | 10.8 | | 2305 |
| Filipino | 59.2 | | | 30.7 | 10.1 | | 2802 |
| Korean | 49.4 | | | 36.0 | 14.6 | | 628 |
| Vietnamese | 56.6 | | | 29.3 | 14.1 | | 663 |
| South Asian | 59.2 | | | 23.1 | 17.7 | | 606 |
| Other Asian** | 58.2 | | | 28.9 | 12.9 | | 973 |
Yin, Liu, Lin, et al [ 10‐ 12] | Chinese (Shanghai) | 50.4 | | | 31.1 | 18.5 | | 4787 |
Xing, Zhao, et al [ 13, 14] | Chinese (Tianjin) | 53.7 | 14.0 | 11.8 | 25.8 | 20.5 | | 3237 |
| Chinese (Taiwan) | 61.8 | 8.8 | 11.8 | 20.5 | 12.8 | 4.9 | 1028 |
Su, et al (this study) | Chinese (Shanghai) | 48.6 | 16.7 | 13.7 | 30.4 | 12.9 | 8.1 | 2791 |
Prognostic significance of breast cancer subtypes among Chinese women
Chinese women with the triple-negative subtype were younger in age at diagnosis compared with women who had other subtypes of breast cancer, which is similar to findings reported in Western populations [
7,
29]. The triple-negative subtype was associated with larger tumor size, higher histologic grade, later TNM stage, and higher prevalence in IDC than in ILC. These clinicopathologic characteristics have been consistently observed in both Western [
4,
8] and Chinese populations [
10‐
13], suggesting that the triple-negative subtype is an aggressive subtype of breast cancer across all ethnicities. Multivariate analysis confirmed that the triple-negative subtype is an independent prognostic factor for the progression and survival of breast cancer. Most triple-negative cancers defined by IHC present a basal-like subtype profile defined by cDNA microarray, but they do not completely correlate in about 25% of cases [
30]. Other molecular subsets may be included in triple-negative cancers. Further epidemiological and biomarker studies for this important subtype in Chinese women is necessary.
The HER2 subtype was closely correlated with larger tumor size and higher histologic grade, consistent with previous reports in other Chinese studies [
10‐
13]. We found that the HER2 subtype was associated with earlier age at diagnosis, more advanced TNM stage, and reduced 5-year overall and disease-free survival rates. Anti-HER2 therapy is currently available. Our study suggests that about 30% of Chinese women with the HER2 subtype (14%) or with the luminal B subtype (17%) may benefit from trastuzumab (Herceptin) and other targeted therapies, if HER2 status were evaluated following the standardized HER2 evaluation guidelines [
15] and this information were incorporated into therapeutic decisions.
The luminal B subtype in our study was correlated with younger age at diagnosis, more advanced TNM stage, larger tumor size, higher histologic grade, and was less common in the ILC and special histologic types compared with the luminal A subtype. However, after adjusting for TNM stage, histologic grade, and tumor size, we observed no statistically significant differences for overall or disease-free survival between the two luminal subtypes. Currently, the definition of the luminal B subtype remains debatable. The luminal B subtype originally classified using cDNA microarray gene profiling was unstable and sometimes clustered with the ER- classes (HER2 and basal subtypes) [
31,
32]. Approximately 30-50% of luminal B class samples defined by gene profiling were HER2+. Therefore, the IHC definition of luminal B (ERα+ and/or PR+, HER2+) is not equivalent to the luminal B tumors classified with microarray gene profiling [
4]. Since the gene profile-classified luminal B subtype is defined as tumors with lower expression levels of ERα/PR and related genes, higher proliferative rates, and higher histologic grade [
32], some authors have suggested that ERα expression in tumor cells should be semi-quantified using the Allred, Q-score, or H-score to distinguish luminal B from luminal A [
33]. More recently, a study [
34] suggested that the Ki67 index for cellular proliferation should be combined with ERα, PR, and HER2 to classify luminal tumors into three subtypes: luminal A (ERα+ and/or PR+, HER2-, Ki67 low), luminal B (ERα+ and/or PR+, HER2-, Ki67 high), and luminal-HER2 (ERα+ and/or PR+, HER2+). In that study, the luminal B and luminal-HER2 subtypes had a statistically significant association with poor breast cancer recurrence-free and disease-specific survival in all adjuvant systemic treatment categories. Additional research is warranted to determine the clinical utility of new methods to distinguish luminal breast cancers.
The immune system is thought to play an important role in the metastatic cascade among cancer patients. Thus, various immune strategies have been tested as therapy for breast cancer, including vaccine therapy, administration of exogenous cytokines, monoclonal antibodies, and gene therapy [
35]. In our study, we collected general information on immunotherapy by asking participants whether they had received immunotherapies such as IL-2, lymphokine-activated killer (LAK) cell, and interferons. We found that use of immunotherapy was associated with improved overall and disease-free survival among women with the luminal A subtype but with reduced disease-free survival among women with the HER2 subtype, suggesting that choosing the proper immunotherapeutic method should be based on the molecular characteristics of the tumor. This also indicates that analysis of molecular subtypes of breast cancer has significance for personalized immunotherapy to improve the survival of breast cancer patients.
In our study, we found that the molecular subtype of breast cancer is not always consistent with histological type in terms of predicting breast cancer outcomes. For example, in our study, medullary breast cancers accounted for about 19% of the HER2 subtype and 38% of triple-negative cases. Medullary breast cancer is generally considered to be a favorable histological type of breast cancer with a good prognosis. The unfavorable molecular subtypes among medullary breast cancer might not mean an unfavorable outcome. These results suggest that breast cancer is more heterogonous than the four molecular subtypes as defined by ER, PR, and HER2 status. Further investigation into molecular heterogeneity is warranted.
This study is the largest population-based study on molecular subtypes of breast cancer and survival among Chinese women. This study has several notable strengths. The population-based study design and high overall response rate (80%) minimized potential selection bias. Standardized staining and scoring of HER2 status, and centralized pathological confirmation of diagnosis minimized misclassification. There are also some limitations to this study. For example, ERα and PR status for the majority of participants (91% and 92%) was obtained from medical charts. Approximately 8% of cases with borderline positivity for HER2 as determined by IHC were not evaluated with FISH. For cases with missing ERα (234 cases) or PR status (222 cases), ER/PR status was measured at the Vanderbilt centralized laboratory using a cut-off for ER positivity of >10%, which is the cut-off used by the large hospitals in Shanghai [
10] and had been validated for the prediction of response to hormonal therapy [
20,
21]. Due to a slightly decreased PR sensitivity of HER2/PR double staining, a lower cut-off positivity value (>1%) was used for PR positivity. To evaluate the potential influence of the variation in the criteria used to define ER and PR status, we performed additional analysis by excluding cases whose ER and PR status were measured at the centralized laboratory. We did not observe appreciable changes in the study results. In 2010, the American Society of Clinical Oncology/College of American Pathologists recommended that ERα and PR should be considered positive, if there are at least 1% positive tumor nuclei in the tissue samples with proper controls [
36]. If the recommended 1% cut-off value for ERα and PR positivity were used in this study, the number of HER2 and triple-negative subtypes would decrease and the number of luminal subtypes would increase. However, the overall prevalence of HER2+ tumors, which includes the luminal B and HER2 subtypes, would not be affected. Future studies on breast cancer subtypes using recommended guidelines [
15,
36] for hormone receptors and HER2 status are warranted. In addition, the follow-up period of this cohort is relatively short. Our ongoing follow-up with the cohort would overcome this limitation and allow an examination of the long-term effects of different molecular subtypes on the survival of breast cancer patients.