The pathologic characteristics of breast cancer in China and its shift during 1999–2008: A national-wide multicenter cross-sectional image over 10 years

Authors

  • Shan Zheng,

    1. Department of Pathology, Cancer Institute & Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, 17 South Panjiayuan Lane, Beijing 100021, China
    Search for more papers by this author
  • Jing-Qiao Bai,

    1. Department of Etiology and Carcinogenesis, State Key Laboratory of Molecular Oncology, Cancer Institute & Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, 17 South Panjiayuan Lane, Beijing 100021, China
    Search for more papers by this author
    • Shan Zheng and Jin-Hu Fan are contributed equally to this work.

  • Jing Li,

    1. Department of Cancer Epidemiology, Cancer Institute & Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, 17 South Panjiayuan Lane, Beijing 100021, China
    Search for more papers by this author
  • Jin-Hu Fan,

    1. Department of Cancer Epidemiology, Cancer Institute & Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, 17 South Panjiayuan Lane, Beijing 100021, China
    Search for more papers by this author
  • Yi Pang,

    1. Department of Cancer Epidemiology, Cancer Institute & Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, 17 South Panjiayuan Lane, Beijing 100021, China
    2. Department of Epidemiology, West China School of Public Health, Sichuan University, Chengdu, Sichuan 610041, China
    Search for more papers by this author
  • Qing-Kun Song,

    1. Department of Cancer Epidemiology, Cancer Institute & Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, 17 South Panjiayuan Lane, Beijing 100021, China
    Search for more papers by this author
  • Rong Huang,

    1. Department of Cancer Epidemiology, Cancer Institute & Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, 17 South Panjiayuan Lane, Beijing 100021, China
    2. Department of Epidemiology, West China School of Public Health, Sichuan University, Chengdu, Sichuan 610041, China
    Search for more papers by this author
  • Hong-Jian Yang,

    1. Department of Breast Surgery, Zhejiang Cancer Hospital, No. 38 Banshanqiao Guanji Road, Hangzhou 310022, China
    Search for more papers by this author
  • Feng Xu,

    1. Department of Breast-thyroid Surgery, Xiangya Second Hospital, Central South University, No. 139 Renminzhonglu, Changsha 410011, China
    Search for more papers by this author
  • Ning Lu,

    Corresponding author
    1. Department of Pathology, Cancer Institute & Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, 17 South Panjiayuan Lane, Beijing 100021, China
    • Department of Pathology, Cancer Institute & Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, P.O. Box 2258, Beijing 100021, People's Republic of China, Tel.: 86 10 8778 8435
    Search for more papers by this author
  • You-Lin Qiao

    Corresponding author
    1. Department of Cancer Epidemiology, Cancer Institute & Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, 17 South Panjiayuan Lane, Beijing 100021, China
    • Department of Cancer Epidemiology, Cancer Institute & Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, P.O. Box 2258, Beijing 100021, People's Republic of China
    Search for more papers by this author
    • Tel.: 86 10 8778 8435/86 10 8778 8489, Fax: 86 10 6771 3648


Abstract

In China, breast cancer is currently the most common malignancy and the sixth leading cause of cancer death in women. But, the characteristics of breast cancer in the whole population are not determined. The aim of this study was to perform a detailed study on pathologic characteristics of breast cancer representing the whole population in China during 1999–2008 and to compare the difference in invasive breast cancer between the Western and Chinese. We randomly collected 4,211 inpatient at seven hospitals in representative geographical regions of China during 1999–2008. All the hospitals had the ability of comprehensive cancer treatment. The pathologic characters including estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) status were surveyed. The shift of pathologic characters was evaluated and the data from China were also compared with those of the Western, both using Chi-square test. We found as follow. (i) The median age of the patients was 48 years and showed the similar characters of Asia. (ii) Breast cancer in China showed more invasive ductal carcinoma with larger tumor size, later stage, lower ER and PR expression and higher HER2 overexpression than those in the Western (p < 0.001). (iii) Both tumor size and stage at diagnosis decreased year by year (p < 0.001). Breast cancer in China showed more aggressive behavior than those in western countries, although tumor size and stage at diagnosis decreased by year during 1999–2008. We addressed the urgent needs for employ race-specific breast cancer screen, diagnosis methods, and therapeutic models in China.

Breast cancer is one of the most common chronic diseases all over the world. It is estimated that more than 1 million women are diagnosed with breast cancer every year, and more than 410,000 will die from it, representing 14.0% of female cancer deaths.1, 2 The incidence of breast cancer was higher in North America, Australia/New Zealand and Europe than that in other countries. It was similar in the age-standard rate for mortality of breast cancer.3 Now, it has become a global issue, as many low- and middle-income countries (LMCs) have reported breast cancer incidence rates increasing by up to 5% per year.4 It is worthwhile that the ratio of the mortality rate to incidence rate is higher in LMCs than that in the developed countries. For example, the ratio is 0.69 in Africa, but it is only 0.19 in North America.5 So, breast cancer is becoming a major health problem in LMCs, due to the late stage at diagnosis, inadequate treatment, pain relief and palliative care.6, 7 Hence, breast cancer research is urgently required in these countries.

Clinical information, diagnostic imaging and pathological characteristics are important factors on the decision of therapy model. Ultra sound and mammography are the most common breast cancer diagnostic imaging techniques, which are essential for early diagnosis. In recent years, breast magnetic resonance imaging (MRI) is widely used in the evaluation of patients with cancer.8–10 Pathological diagnosis is one of the important components for breast cancer therapy. The information of tumor size, histological type, biomarker status, and TNM stage were the essential components for therapy.11 In recent years, estrogen receptor (ER), progesterone receptor (PR) and human epidermal growth factor receptor 2 (HER2) have been used for tumor subtype classification and treatment stratification.3, 12, 13 ER and PR are ligand-activated transcription factors belonging to the nuclear hormone receptors family.14 It has been demonstrated that ER-positive (ER+) and/or PR-positive (PR+) patients can benefit from adjuvant endocrine treatment.12, 15 HER2 protein is a member of subclass 1 of the receptor tyrosine kinases superfamily.16 Patients with HER2 amplification or overexpression generally respond to Trastuzumab-based target therapy.17 In summary, rationalized application of diagnosis and treatment are important in improving the prognosis of breast cancer patients.

In China, breast cancer is currently the most common malignancy and the sixth leading cause of cancer death in women. The incidence and mortality rates show a rising trend, and present significant regional differences.18, 19 Although extensive work has been done to study Chinese women with breast cancer,20–26 all the previous studies only focus on population in northern or eastern of China and the majority studies are single-center/hospital-based. No study has been reported to characterize breast cancer of the whole population in China. The aim of this study was to perform a detailed study on pathologic characteristics of the breast cancer representing the whole population in China during 1999–2008. At the same time, pathologic characteristics of invasive breast cancer between the Western and Chinese were also compared. As imaging, including ultra sound, mammography and MRI, and pathological characters including ER, PR and HER2 status are the major factors influence cancer diagnosis and subgrouping, these information on each individual are collected for this study.

Abbreviations

AJCC: American Joint Committee on Cancer; BCCOM: Breast Cancer Clinical Outcome Measures; BHGI: Breast Health Global Initiative; BMI: body mass index; BRCA1: breast cancer susceptibility gene 1; C.I.: confidence interval; CIS: carcinoma in situ; CRF: case report forms; DCIS: ductal carcinoma in situ; ER: estrogen receptor; HER2: human epidermal growth factor receptor 2; IDC: invasive ductal carcinoma; IHC: immunohistochemistry; ILC: invasive lobular carcinoma; LCIS: lobular carcinoma in situ; LMC: low- and middle-income countries; LNM: lymph node metastasis; MRI: magnetic resonance imaging; −; Negative; OR: odds ratio; +: Positive; PR: progesterone receptor

Materials and Methods

Study design and quality control

This study was a multicenter retrospective hospital-based research focused on Chinese women with primary breast cancer, which had passed the review of the Ethics Committee of Cancer Foundation of China. Patient consent was not required for this study as there were no risks anticipated to the participants of the study. The data were stripped of any patient identifiers per the approved procedures. De-identified data were maintained on secure database. Only research team members have access to the data. All data will be reported in aggregate.

One hospital with good standard quality from each of the seven China districts, north, northeast, northwest, central, east, south and southwest of China, which showed different levels of breast cancer burden,19 were included in this study. The hospitals were selected according to their abilities to integrate the diagnosis and properly treat the patients and, most importantly, that they could represent for the regional patients resource. There were seven tertiary hospitals selected in this study. All patients enrolled in this study must meet three key inclusion criteria: (i) pathology confirmed primary breast cancer; (ii) inpatient admission date was within the selected month in the study hospital and (iii) received or receiving treatment (surgery, medical oncology and radiotherapy) for breast cancer through a site. Each hospital collected data randomly for no less than 50 cases in any month from March to December during 1999–2008, so that each hospital offered a total of no less than 500 cases in these 10 years. In total, data from 4,211 cases were collected with quality control, including proper case report forms to extract the information, local clerk training, double-enter of data, validation by EpiData and consistency check. The methodology was described in detail in a separate report.27

Data collection and stratification

Primary medical reports were extracted, including general information, risk factors, diagnostic imaging test, therapeutic model and pathologic characteristics. Histological subtype was referred to 1981 and 2003 WHO histological classification criterion.28, 29 Staging of breast cancer was referred to 1997 and after American Joint Committee on Cancer (AJCC) TNM stage criterion.30, 31 The expression of ER, PR and HER2 were detected by immunohistochemistry using ER (Clone: SP1, Neomarkers), PR (Clone: SP2, Neomarkers) and HER2 (Clone: A0485, Dako) antibodies. ER, PR, and HER2 status of a breast cancer were primarily extracted from medical records. In this study, patients were considered HER2+, that is, with overexpression of HER2, if it was scored as 3+; patients with HER2 scored as 0, 1+, 2+ were considered HER2 negative (HER2-).13 According to guideline in diagnostic resource from Breast Health Global Initiative Consensus Conference,8 we evaluate the diagnostic imaging resource allocation each year into four levels: basic, limited, enhanced and maximal. In brief, at basic level, diagnosis is made on clinical history and clinical breast examination without any imaging services, while breast ultra sound is added to diagnose at limited level. With enhanced-level, mammography is usually applied to diagnosis with or without ultra sound. With maximal-level, significant cost imaging services are used for diagnosis with or without ultra sound or mammography, such as breast MRI.

Comparison of multiple group data

Frequently cited pathologic data from US and Europe, such as the US Surveillance, Epidemiology and End Results (SEER) database, Breast Cancer Clinical Outcome Measures (BCCOM) database, the European ONCOPOOL database, and other large-scale population studies,32–38 were used to compare the characteristics between the Western and Chinese. In detail, the Caucasian data of tumor size, pathologic type, AJCC stage, ER and PR status from US SEER database and HER2 status data from the IMPATH laboratories were used as US representative data.32, 34, 37 Tumor size data from BCCOM database,33 pathologic type from ONCOPOOL database,35 AJCC stage, ER and PR status from Switzerland36, HER2 status from England38 were used as Europe representative data.

Statistical analysis

Chi-square tests were performed to evaluate the shift of tumor size, lymph node metastasis (LNM), pathologic type, tumor stage, the biomarker expression of breast cancer between the two periods of 1999–2003 and 2004–2008. In addition, Chi-square tests were used to compare these corresponding characters of invasive breast cancer between the Western and Chinese. Categorical variables were analyzed by Spearman Chi-square test. The ordinal variables more than two levels were analyzed by Cochran-Armitage test for trend. The time trend of each pathologic character of invasive breast cancer was evaluated by Chi-square test. The trend of ER, PR and HER2 status were estimated by Cochrane-Armitage trend tests. The temporal trend of tumor size, pathological types, LNM and tumor stage were measured by Mantel–Haenszel Chi-square tests. Binary Logistic regression was performed to evaluate the relationship between the dependent factors and covariates. The dependent factors included tumor size, LNM, pathologic type, tumor stage, ER, PR and HER2 status. All these dependent factors were divided into two levels when we evaluated their time trend. In detail, tumor size was divided into ≤2 and >2 cm levels, and that LNM, ER, PR and HER2 were all divided into “+” and “-” levels. Tumor stages were divided into stage I and stage II–IV levels. As for pathologic type, we focused on the shift of the ratio of invasive ductal carcinoma (IDC) to other invasive carcinoma than IDC. The covariates included year, diagnostic imaging resource allocation, age and menstrual status of patient. Diagnostic imaging resource allocation was divided into four levels.8 Menstrual status was divided into two groups of premenopausal and postmenopausal. Only the pathologic characters which showed time trend in Cochrane–Armitage trend tests or Mantel–Haenszel Chi-square tests were included in Binary Logistic regression. For limited sample size of carcinoma in situ (CIS), the trend of pathological characters was not estimated. All the statistical analyses were performed using the software SPSS 16.0 (SPSS Chicago, IL). Statistical significance was assessed by two-tailed tests with p < 0.05.

Results

A total of 4,211 cases were collected during 1999–2008, accounting for 9.3% of the total breast cancer cases during 1999–2008.27 Median age at diagnosis was 48 ranging from 21 to 86-year-old. There were 2,649 (62.9%) premenopausal patients and 1,562 (37.1%) postmenopausal patients. All other information including age distribution, education level, risk factor exposures and therapeutic model was described in detail in a separate report.27 In this article, we focused on diagnostic imaging utilization, tumor size, LNM, pathologic type, tumor stage and biomarker expression. We divided tumor size into three groups: ≤2, >2–5 and >5 when we described the pathologic characters of breast cancer in China.

The information of pathologic characters

There were 3,856 (3,856/4,211, 91.6%) patients diagnosed as invasive breast cancer with the main pathological type as IDC (3,471/3,856, 90.0%). Tumor size ≤ 2cm accounted for 25.9% (997/3,856) and >2–5cm alone for 51.3% (1,977/3,856). Stage I cases accounted for 16.4% (632/3,856) and stage II cases alone of 47.3% (1,823/3,856). The proportion of LNM positive was 48.3% (1,863/3,856). The ER+ rate was 49.5% (1,909/3,856), while PR+ rate was 50.1% (1,931/3,856). The HER2 + rate was 18.0% (694/3,856). Table 1 summarized the information of pathologic characters in invasive breast cancer. There were 143 (143/4,211, 3.4%) patients diagnosed only with CIS of breast with the main pathological type of ductal CIS (133/143, 93.0%). Lesion size ≤ 2cm accounted for 37.8% (54/143) and >2–5cm alone of 35.0% (50/143). The ER+ rate was 51.7% (74/143), while PR+ rate was 55.2% (79/143) (Supporting Information, Table S1). The pathological types cannot be assigned for 212 (212/4,211, 5.0%) patients.

Table 1. Pathologic characters and their shift of invasive breast cancer during 1999–2008 (n = 3856)
inline image

Trends for diagnostic imaging resource allocation

Breast ultrasound and mammography were the commonly used diagnostic imaging methods for breast cancer. In recent years, MRI has been increasingly used (Supporting Information, Table S2). Figure 1 summarized the diagnosis imaging resource allocation and its trend during 1999–2008. We found significant difference in annual diagnostic resource level between periods from 2004–2008 to 1999–2003 (Supporting Information, Table S3). Therefore, we divided the patients into two groups, 1999–2003 and 2004–2008 for the primarily comparison of the pathologic characters.

Figure 1.

Trends in diagnostic imaging utilization during 1999–2008. Solid black line represents “basic level.” Dotted gray line with squares represents “limited level.” Dotted gray line with triangles represents “enhanced level.” Dotted gray line with cross signs represents “maximal level.” Limited level was the main diagnosis imaging resource allocation level in China during these 10 years. The ratio of basic level reached peak at 2002 and decreased due to the increase of enhanced level since 2003. The ratio of maximal level increased after 2005. Unknown cases are not shown in the figure.

Shifts in pathologic characters

Compared with 1999–2003 group, tumor size of invasive breast cancer in 2004–2008 group was significantly reduced (p < 0.001). The proportion of IDC increased while other special types of invasive breast cancer decreased (p = 0.01). The proportion of stage I tumors increased while the total proportion of more than stage I decreased (p < 0.001). The rate of ER+ (p = 0.03), PR+ (p = 0.02) increased in 2004–2008 group. HER2+ showed some changes, but did not reach statistical significance (p = 0.059). LNM also showed some changes, but without statistical significance (p = 0.551). Detailed information for these statistical analyses is shown in Supporting Information Table S4. Supporting Information Figure S1 summarized the shift of pathologic characters in invasive breast cancer between 1999–2003 and 2004–2008 groups. The pathologic characters in CIS of breast also showed some shift, but none of them were statistical significant (p < 0.05). Then, we analyzed the time trend of invasive breast cancer during 1999–2008 (Table 1). Tumor size (p < 0.001), pathological type (p = 0.023) and AJCC stage (p = 0.009) of breast cancer all had significant changing trend. The proportion of other factors, including LNM, ER, PR and HER2 status showed some changes, but did not reach statistical significance (p > 0.05). At last, we applied Binary Logistic regression to analyze the data thoroughly, taking into account of year, age, menstrual status and diagnostic imaging resource allocation of patients. Only tumor size, pathological type and tumor stage were included in Binary Logistic regression. As age and menstrual status were obvious correlative, we took year, diagnostic imaging resource allocation and age or menstrual status of patients as covariates in Binary Logistic regression analysis. As shown in Table 2 and Figure 2, both tumor size and stage had decreased year by year (p < 0.001), while pathologic type showed some changes, but did not reach statistical significance (p > 0.05).

Figure 2.

Changes in tumor size and stage of invasive breast cancer during 1999–2008. Solid black line with triangles represents “tumor size ≤2 cm.” Solid black line with squares represents “tumor size >2cm.” Dotted gray line with triangles represents “stage I.” Dotted gray line with squares represents “stage II–IV.” Both tumor size and tumor stage show decrease trend. Unknown cases are not shown in the figure.

Table 2. The relationship between tumor size, tumor stage of invasive breast cancer in China and year, diagnostic imaging resource allocation age or menstrual status
inline image

Differences between the Western and Chinese

The features of invasive breast cancer in the Western and Chinese population were also compared. As shown in Figure 3 and Supporting Information Table S5, invasive breast cancer in China showed larger tumor, more IDC, later stage, lower ER+ and PR+, and higher HER2+ (p < 0.001) than those of the Western population, although the 2004–2008 trend of all the characters except HER2 were going to the Western direction compared with the 1999–2003 data. Our results indicated that invasive breast cancer of Chinese may be more aggressive than those of the Western population.

Figure 3.

Comparison of characters of invasive breast cancer between the Western and Chinese population. White bars represent characters data in US. Striped bars represent characters data in Europe. Black bars represent characters data in China from 1999 to 2003. Gray bars represent characters data in China from 2004 to 2008. *significant difference <0.001. Unknown cases are not shown in the figure.

Discussion

Breast cancer has become a global healthcare problem. Although the incidence rate is high in developed country, it is more harmful in LMCs.5 As showed in the survey of the dead cause in China, breast cancer mortality rate increased by nearly 100% during the past three decades.39 The study showed the image of pathologic characters of female breast cancer covering seven traditional districts in China during 1999–2008.

In other Asian countries, IDC was the main pathological types of breast cancer (73.4–90.0%), and tumor size ranged from 2.0 to 5.0 cm (48.6–58.3%). Stage II was very common (43.5–58.3%), while the proportion of stage I was low (3.4–24.2%). A lot of patients showed LNM positive (36.9–70.2%). The rate of ER+ and PR+ ranged from 16.1 to 57.5% while the rate of HER2+ ranged from 14.5 to 30.1%.23, 40–47 Most of our data were at the intermediate levels of these ranges, showing that the characters of Chinese are similar to that of other Asian countries. The similarities of Chinese population and other Asian countries may reflect the similarity in genetic background. Moreover, our results may also reflect that the healthcare resource of breast cancer in China was at a medium level in Asia.

Chinese population is a low-risk population of breast cancer, where, however, incidence rates are increasing rapidly. Figure 4 showed the comparison of incidence rate in North America, Europe and different districts all over China, according to global cancer statistics in 2002 and Chinese cancer registry annual report in 1998–2002.1, 48 Here, world age standard incidence rates of North America, Europe and different districts all over China were listed. The incidence rate in North America and Europe were more than 60.0 per 100,000, while the incidence rate in most districts of China was lower than 30.0 per 100,000. In this study, we wanted to confirm whether there were some differences in pathologic characteristic between the Western and Chinese population. As invasive breast cancer was the main pathologic type all over the world, we only compared the features of invasive breast cancer between the Western and Chinese population. We selected US as the representative population of North America. The sources of the pathologic characteristic data from the Western are heterogeneous and we took into account variability in these estimates. The patterns from different western studies are similar, which can be grouped together, but the Chinese pattern is different from the others. Breast cancer in Chinese population showed younger age at diagnosis, larger tumor size, later tumor stage, a higher proportion of IDC, lower rate of ER+ and PR+ as well as higher rate of HER2+ than those in the Western population, although a few parameters were going toward the Western characters in recent years. In addition to the difference of constitutive ratio of age, life style, socio-economic level, education level, reproductive factors of patients between the two populations, the difference reflects the differences in healthcare resource level between the Western and Chinese population.23, 40, 41 The lack of breast cancer screening in China should reduce the chance of early diagnosis, which is opposite to the observed younger age at diagnosis in China compared to the Western countries. Apart from the above environmental and social economic factors, the differences of pathological features and biological markers between the Western and Chinese population may also reflect the magnitude of background breast cancer risk,1 the differences in genetic background such as estrogen receptor alpha gene polymorphisms.49 In prostate cancer, it has recently been revealed that different genomic alterations exist in Chinese and the Western cancers.50 However, there are few studies to be investigated if there is a genomic alteration difference between the Chinese and Western breast cancers. Moreover, our results suggest that Chinese women need breast cancer screening for earlier age and that Chinese invasive breast cancer may be more aggressive than the Western counterpart and need more specific therapeutic models. However, we only compared crudely the characteristics between the Western and Chinese population without eliminating the effects of age as the individual age data from the Western studies are not available. We provided our data in age breakup groups (Supporting Information, Table S6) for other researchers to do age justified analysis using our data.

Figure 4.

The comparison of incidence rate in North America, Europe and different districts all over China, according to global cancer statistics (2002) and Chinese cancer registry annual report (1998–2002). The world age standard incidence rate of North America, Europe and the seven districts all over China were listed. The light gray bar shows the incidence rate lower than 30.0 per 100,000. The heavy gray bar shows the incidence rate ranged from 30.0 to 80.0 per 100,000. The black bar shows the incidence rate higher than 80.0 per 100,000. The incidence rate in North America and Europe is much higher than that in China. The incidence in different areas of China ranges from 5.2 to 31.2 per 100,000.

In this study, we identified the shift of these characters during 1999–2008. According to the size of sample, only the pathologic characters of invasive breast cancer were evaluated. We first analyzed our data primarily by dividing these 10 years into two periods of 1999–2003 and 2004–2008 based on three reasons. First, this could reduce the effects of variations between different years. Second, there are significant differences between the two groups in annual diagnostic resource level, which is essential for early detection of breast cancer. Third, Chinese government had been gradually amending the policy of health insurance since 2004. More and more elder patients went to hospitals to treat their diseases including malignant tumor, which may change the constitutive ratio of age at diagnosis of patients (data not shown). We observed significant changes of the pathologic characters between the two groups of 1999–2003 and 2004–2008. Then, we analyzed the data of invasive breast cancer year by year to find the potential reasons for the shift. We took two steps. First, we analyzed the time trend of each pathologic character by Cochrane–Armitage trend tests or Mantel–Haenszel Chi-square tests according to the type of pathologic character. The results showed that tumor size, pathologic type and tumor stage showed significant changing trend (p < 0.05). Second, we use Binary Logistic regression to evaluate the positive pathologic character in Chi-square trend test, taking into account of age, menstrual status and diagnostic imaging resource allocation. The pathologic character included tumor size, pathologic type and tumor stage. We wanted to confirm whether the result of time trend show the actual fact or it is only the reflection of other covariates such as age, menstrual status and diagnostic imaging resource allocation. The results showed that only tumor size and tumor stage presented decreasing trend by year. As shown in Supporting Information Table S7, the proportion of educational level in secondary high school or higher had been lifted year by year (p < 0.001), while age at first birth, age at menopause, body mass index and the proportion of breastfeeding of the patients in this study showed little shift during 1999–2008 (p > 0.05). The former may partly promote the increase in the small and early stage tumor of invasive breast cancer at diagnose by year. The higher education level people received, the stronger consciousness of health care they got. Higher educated people always went to hospital for help in early disease condition. The gradual increase of GDP in China during 1999–2008 may also change the life style. People had more approach to acquire the knowledge of health education from internet, television, newspaper as well as traditional textbook. This may also partly promote the increase in the small and early stage tumor of invasive breast cancer by year. However, this shift may also reflect the difference background levels of risk in breast cancer patients between urban and rural areas.

There are some potential limits in this study. We selected patients from seven tertiary hospitals, which were among the best hospitals in each district, to represent the characters of the whole population. Selection bias may exist in the catchment of breast cancer patients in the selected hospital. However, the selection of these tertiary hospitals is based on the consideration of their ability in keeping full patient clinicopathological records, and most importantly to guarantee the national standard of quality control in laboratory assays, therefore minimize inter-hospital variations. All these hospitals also have the abilities of comprehensive therapy of cancers.

In conclusion, breast cancer in China showed similar characters of Asia in tumor size, stage, pathologic type and positive rate of LNM, ER, PR and HER2 status. However, these characters are significantly different from those of the Western population, although a few parameters such as tumor size and stage at diagnosis are going toward the Western characters in recent years. According to pathologic characters, breast cancer in China showed more aggressive behavior than those in western countries. Our study provides new findings supporting breast cancer can vary between different ethnic populations due to the magnitude of background breast cancer risk, genetic and/or environment discrepancies. We also address the urgent needs for employ race-specific breast cancer screen, diagnosis methods and therapeutic models in China.

Acknowledgements

The authors thank Cancer Foundation of China originating this retrospective clinical epidemiological study of breast cancer. They thank the local investigators from Beijing, Liaoning (Shenyang), Hunan (Changsha), Guangdong (Guangzhou), Zhejiang (Hangzhou), Shanxi (Xian) and Sichuan (Chengdu) for data collection and assisting us complete the project successfully. The authors also thank Pfizer for funding in the form of donations to Cancer Foundation of China.

Ancillary