A population-based case-control study on the association of Angelica sinensis exposure with risk of breast cancer.

In Taiwan, national health insurance coverage rate is over 99.9%, and its health insurance information known as national health insurance research database (NHIRD) is a representative empirical dataset in the field of health care related research. 10 We conducted a population-based case-control study design based on Taiwan’s National Health Insurance claim data (medical claims of all cancer patients and of a random sample of 1-million people registered in 2005) released by the National Health Research Institutes. The case series was all newly diagnosed breast cancer patients with the International Classification of Disease 9th version Clinical Modification (ICD-9-CM) codes 174.X and catastrophic illness registration in Taiwan between 2005-2008; the control group was randomly selected from the 1-million beneficiaries who registered with the National Health Insurance program in 2005 and had no breast cancer diagnosis between 2000 and 2008. To improve the comparability, we employed the time density sampling method to select controls which were matched to case on date of breast cancer diagnosis and age in year, with a case/control ratio of 1/3. The prescriptions of the Angelica sinensis in Taiwan are carried out by the TCM doctors, qualified by the national examination, with the supervision of the Health Insurance Bureau. Both Angelica sinensis individual and compound drugs used are scientific Chinese medicine powders produced by GMP pharmaceutical factories, and often applied to the syndromes of blood deficiency and blood stasis determined by TCM doctors, with mean dosage generally single herb 1.5 g/day, compounds are 4–6g/day. Exposure to Angelica sinensis was defined by the use of single prescription of Angelica sinensis or compounds in which the formula proportion of Angelica sinensis was greater than or close to 10% during the observation period. In addition, the commonly-used formulas containing Angelica sinensis surveyed by previous studies were also included. 7 Exposure of Angelica sinensis and other potential confounders was retrospectively retrieved between 2000/1/1 and date of breast cancer diagnosis. Both exogenous hormone and selected co-morbidity were considered as potential confounders in this study. Exposure to exogenous hormone was defined as ever using medication containing estrogen and progesterone before the diagnosis of breast cancer. The diagnosis of malignant neoplasm of female genital organs (include malignant neoplasm of ovaries, uterus, and cervix uteri) was according to ICD-9-CM codes 180.X, 182.X, 183.X, 184.X; benign neoplasm of breast, benign uterine tumor (include endometriosis, leiomyoma of uterus) and metabolic disease (include obesity, hypertension, disorders of lipoid metabolism) were determined based on ICD-9-CM codes 217, 617.X, 218.X, 278.X, 401.X, and 272.X, respectively. The age of case group was set at the time being diagnosed with breast cancer and control group was set at the time being matched. The insurance payment and areas were captured according to the registry for beneficiaries in 2000 as baseline of claim data. Information of the above-mentioned potential confounders were identified from both inpatient and outpatient between 1997/1/1 and the diagnosis of breast cancer. We analyzed all data with SAS (version 9.4; SAS Institute, Cary, NC). Descriptive statistics and analytical statistics were employed conforming to the study purpose and the variable property. The α level was set at 0.05. It was used to describe and compare between case and control group with respect to various socio-demographic variables, medications, and co-morbidity, including age, insurance premium based salary, residential areas, diagnosis of gynecologic cancer, benign cyst of breast, benign neoplasm of uterus, disorder of metabolism, exposure history of estrogen and progesterone. We calculated means and standard deviations for continuous variables, as well as number and percentage for categorical variables. . We used simple conditional logistic regression model was used to estimate the crude odds ratio (OR) of breast cancer in association with exposure to both single and compound prescriptions of Angelica sinensis . The adjusted OR of breast cancer was further estimated from multiple conditional logistic regression by taking into account the potential confounders. The trend test was used to observe whether the dose-response relationship exist between exposure to Angelica sinensis and breast cancer. We also assessed the time period between initial use of Angelica sinensis and breast cancer diagnosis in association with breast cancer risk; and explored whether age (≦47 years, >47-<55 years, and ≧55 years) at first exposure to Angelica sinensis may pose differential influences on breast cancer risk. The later analysis was set to investigate the potential influence of menopause on the relation between Angelica sinensis and breast. We performed two sensitivity analyses to assess the potential confounding bias that could be involved in this study. First, there has been concern since 2005 over the potential breast cancer risk associated with Angelica sinensis use. We therefore performed separate analyses based on the data before and after 2005 to assess the potential confounding by indication, in which TCM doctors might tend to not prescribe Angelica sinensis to women at potentially higher risk of breast cancer (e.g., abnormal mammography report or family history of breast cancer) after 2005. Second, we also used acupuncture habit as a negative exposure, which is believed to have no effect on breast cancer incidence, to assess the potential influence of unmeasured confounders (mainly socioeconomic status and reproductive factors).

The present study included total number of 34,262 cases and 102,786 controls. The average age of both groups was similar at around 53 years old. More cases than controls were living in Northern areas (more urbanized) and having higher salary. The prevalence of phytohormone exposure and selected co-mobility was also higher in cases than in controls ( Table 1 ). Table 1 Demographics and clinical characteristics of cases and controls. Table 1 Demographics and clinical characteristics Cases N = 34262 Controls N = 102786 Crude OR (95% CI) P Model 1 P Model 2 Mean (SD) or No. (%) Mean (SD) or No. (%) Adjusted OR (95% CI) Adjusted OR (95% CI) P Age, years 53.2 (12.2) 53.1 (12.2) 1.00 0.53 Residential area North 18055 (54.6) 49762 (49.5) 1.00 (REF) 1.00 (REF) Central 5348 (16.2) 17735 (17.6) 0.83 (0.80–0.86) <.0001 0.97 (0.94–1.01) 0.19 South 8790 (26.6) 30043 (29.9) 0.81 (0.78–0.83) <.0001 0.89 (0.86–0.93) <.0001 East 693 (2.1) 2387 (2.4) 0.80 (0.73–0.87) <.0001 0.93 (0.84–1.02) 0.13 Islands 194 (0.6) 596 (0.6) 0.90 (0.76–1.06) 0.19 1.02 (0.85–1.22) 0.84 Insurance premium based monthly salary (NT$) Dependent 8483 (25.5) 24890 (24.8) 1.00 (REF) 1.00 (REF) 1-19,999 14226 (42.7) 50222 (50.0) 0.83 (0.80–0.86) <.0001 0.86 (0.83–0.89) 39,999 3932 (11.8) 6570 (6.5) 1.76 (1.67–1.84) <.0001 1.60 (1.51–1.69) <.0001 Gynecology cancer 711 (2.1) 1376 (1.3) 1.55 (1.41–1.70) <.0001 1.57 (1.42–1.74) <.0001 1.57 (1.41–1.74) <.0001 Benign breast tumor 14896 (43.5) 8399 (8.2) 8.65 (8.39–8.93) <.0001 8.79 (8.49–9.10) <.0001 8.65 (8.35–8.96) <.0001 Benign uterine tumor 6192 (18.0) 15069 (14.7) 1.29 (1.25–1.33) <.0001 1.05 (1.01–1.09) 0.02 1.04 (1.00–1.09) 0.03 Metabolic disease 14904 (43.5) 42642 (41.5) 1.09 (1.07–1.12) <.0001 1.02 (0.99–1.05) 0.20 1.06 (1.03–1.10) <.001 Estrogen exposure 395 (1.2) 1206 (1.2) 0.98 (0.88–1.10) 0.76 0.95 (0.83–1.08) 0.43 0.97 (0.85–1.11) 0.70 Progesterone exposure 1175 (3.4) 3718 (3.6) 0.95 (0.89–1.02) 0.14 0.88 (0.82–0.96) <0.01 0.88 (0.81–0.95) <0.01 Abbreviations: SD, standard deviation; CI, confidence interval; OR, odds ratio; NT$, New Taiwan dollar. Model 1 adjusted only co-morbidity and estrogen/progesterone exposures. Model 2 further adjusted demographic characteristics in addition to the factors adjusted in Model 1. Demographics and clinical characteristics of cases and controls. Abbreviations: SD, standard deviation; CI, confidence interval; OR, odds ratio; NT$, New Taiwan dollar. Model 1 adjusted only co-morbidity and estrogen/progesterone exposures. Model 2 further adjusted demographic characteristics in addition to the factors adjusted in Model 1. The single herb prescription of Angelica sinensis was accounted for 6.2% and 6.3% in case and control group respectively, which was much less than the compound prescription accounting for 49.9% and 49.8%. This has shown the usage of Angelica sinensis in Taiwan was compound prescription in large degree, corresponding to the literature review previously mentioned. 7 The total prevalence of Angelica sinensis use is similar for cases (50.7%) and controls (50.6%), representing a crude OR of 1.02 (95% confidence interval (CI) 0.99–1.04, P value = 0.24). After controlling for potential confounders, the adjusted OR was reduced to 0.93(95% CI 0.90–0.96, P value < 0.0001), revealing that the use of Angelica sinensis and breast cancer risk manifested a weak but statistically significant protective effect( Table 2 ). Analysis of the dose-response relationship between use of single and compound prescription of Angelica sinensis and breast cancer risk showed an adjusted OR of 0.95 (95% CI 0.92–0.98, P value < 0.01), 0.92 (95% CI 0.88–0.95, P value < 0.0001), and 0.91 (95% CI 0.87–0.96, P value < 0.0001)for the accumulated dose of 0.1–9.9 g, 10–29.9 g, and more than 30 g, respectively, with a significant downward trend (beta = -0.024, P value < 0.001). ( Table 2 ). Table 2 Crude and adjusted ORs of breast cancer in association with Angelica sinensis exposure. Table 2 Cases N = 34262 Controls N = 102786 Crude OR (95% CI) P Adjusted OR (95% CI) P No. (%) No. (%) Angelica sinensis exposure No 16879 (49.3) 50767 (49.4) 1.00 (REF) 1.00 (REF) Yes 17383 (50.7) 52019 (50.6) 1.02 (0.99–1.04) 0.24 0.95 (0.93–0.98) <0.0001 Exposure dose (grams) 0.1–9.9 8017 (23.4) 24271 (23.6) 1.01 (0.97–1.04) 0.76 0.97 (0.93–1.00) <0.06 10–29.9 4917 (14.4) 14827 (14.4) 1.01 (0.97–1.05) 0.69 0.93 (0.89–0.97) <0.001 ≧30 4039 (11.8) 11612 (11.3) 1.06 (1.02–1.10) <0.01 0.96 (0.91–1.00) <0.06 Trend test: β=-0.024 <0.001 Abbreviations: CI, confidence interval; OR, odds ratio. Adjusted OR: adjusted for residential area, monthly salary, gynecology cancer, benign breast/uterine tumor, metabolic disease and estrogen/progesterone exposure. Note: the total exposure dose contain Angelica sinensis compound dose x0.1 and Angelica sinensis single herb dose x1. Crude and adjusted ORs of breast cancer in association with Angelica sinensis exposure. Abbreviations: CI, confidence interval; OR, odds ratio. Adjusted OR: adjusted for residential area, monthly salary, gynecology cancer, benign breast/uterine tumor, metabolic disease and estrogen/progesterone exposure. Note: the total exposure dose contain Angelica sinensis compound dose x0.1 and Angelica sinensis single herb dose x1. Table 3 shows the ORs of breast cancer in relation to the time period between initial use of Angelica sinensis and breast cancer diagnosis. The risk of breast cancer was not significantly associated with a time period less than 6 years. However, when the initial use of Angelica sinensis was more than 6 years before breast cancer diagnosis, the significantly protective effect appeared. In addition, there is a tendency that the earlier the initial use of Angelica sinensis happened, the stronger the seemingly protective effect of Angelica sinensis was observed. The regression coefficient of trend test was -0.011 (P value <0.0001). ( Table 3 ). Table 3 Crude and adjusted ORs of breast cancer in association with time period between initial use of Angelica sinensis and breast cancer diagnosis. Table 3 Cases N = 34262 Controls N = 102786 Crude OR (95% CI) P Adjusted OR (95% CI) P No. (%) No. (%) Non-exposure 16879 (49.3) 50767 (49.4) 1.00 (REF) 1.00 (REF) Time period between initial use and breast cancer diagnosis 0y< ∼ ≦1y 1342 (3.9) 3584 (3.5) 1.12 (1.05–1.20) <0.01 1.08 (1.00–1.16) 0.06 1y< ∼ ≦2y 1535 (4.5) 4575 (4.5) 1.03 (0.96–1.09) 0.43 0.95 (0.89–1.02) 0.16 2y< ∼ ≦3y 1744 (5.1) 5281 (5.1) 1.00 (0.95–1.06) 0.98 0.96 (0.90–1.03) 0.22 3y< ∼ ≦4y 2085 (6.1) 6342 (6.2) 1.00 (0.95–1.06) 0.99 0.95 (0.90–1.01) 0.10 4y< ∼ ≦5y 2670 (7.8) 7833 (7.6) 1.04 (0.99–1.09) 0.14 0.98 (0.92–1.03) 0.37 5y< ∼ ≦6y 2956 (8.6) 8816 (8.6) 1.02 (0.97–1.07) 0.41 0.96 (0.91–1.01) 0.14 6y< ∼ ≦7y 2410 (7.0) 7383 (7.2) 0.99 (0.94–1.04) 0.71 0.91 (0.86–0.97) <.01 7y< ∼ ≦8y 1761 (5.1) 5573 (5.4) 0.96 (0.91–1.02) 0.16 0.87 (0.82–0.94) <.0001 8y< ∼ ≦9y 880 (2.6) 2631 (2.6) 1.02 (0.94–1.10) 0.62 0.96 (0.87–1.06) 0.41 Trend test: β=-0.011 <.0001 Abbreviations: CI, confidence interval; OR, odds ratio; y, years. Adjusted OR: adjusted for residential area, monthly salary, gynecology cancer, benign breast/uterine tumor, metabolic disease and estrogen/progesterone exposure. Crude and adjusted ORs of breast cancer in association with time period between initial use of Angelica sinensis and breast cancer diagnosis. Abbreviations: CI, confidence interval; OR, odds ratio; y, years. Adjusted OR: adjusted for residential area, monthly salary, gynecology cancer, benign breast/uterine tumor, metabolic disease and estrogen/progesterone exposure. This study stratified women according to age at exposure to Angelica sinensis . The most obvious protective effect was in women whose initial Angelica sinensis exposure was at prei-menopausal ages (47–55 years) (aOR: 0.93, 95% CI 0.88–0.98). The adjusted OR was marginally significant for exposure at premenopausal ages, but was insignificant for exposure to Angelica sinensis after menopausal ages ( Table 4 ). Table 4 Crude and adjusted ORs of breast cancer in association with initial use of Angelica sinensis at various ages. Table 4 Age at initial use of Agelica sinensis Cases N = 34262 Controls N = 102786 Crude OR (95% CI) P Adjusted OR (95% CI) P No. (%) No. (%) Non-exposure 16879 (49.3) 50767 (49.4) 1.00 (REF) 1.00 (REF) Exposure before menopause (age≦47y) 9119 (26.6) 27152 (26.4) 1.02 (0.99–1.05) 0.17 0.96 (0.92–1.00) 0.04 Exposure during menopause (47y < age<55y) 4152 (12.1) 12614 (12.3) 1.00 (0.96–1.04) 0.95 0.93 (0.88–0.98) <.001 Exposure after menopause (age≧55y) 4112 (12.0) 12253 (11.9) 1.02 (0.98–1.06) 0.40 0.97 (0.92–1.02) 0.25 Abbreviations: CI, confidence interval; OR, odds ratio; y, years old. Adjusted OR: adjusted for age, residential area, monthly salary, gynecology cancer, benign breast/uterine tumor, metabolic disease and estrogen/progesterone exposure. Crude and adjusted ORs of breast cancer in association with initial use of Angelica sinensis at various ages. Abbreviations: CI, confidence interval; OR, odds ratio; y, years old. Adjusted OR: adjusted for age, residential area, monthly salary, gynecology cancer, benign breast/uterine tumor, metabolic disease and estrogen/progesterone exposure. In the first sensitivity analysis, we noted that Angelica sinensis exposure before 2005 was associated with a significantly reduced risk of breast cancer with an aOR of 0.94 (95% CI 0.92-0.97). The aOR associated with after Angelica sinensis exposure after 2005, on the other hand, showed no significantly lower aOR (1.02, 95% CI 0.96-1.08). The above results showed no obvious confounding by indication for our data. The second sensitivity analysis revealed that exposure to acupuncture was not significantly associated with breast cancer with an aOR of 1.05 (95% CI 1.01–1.09). In addition, the aOR associated with 1–4, 5–9, and ≥10 times of acupuncture use was estimated at 1.03 (95% CI 0.98–1.08), 1.06 (95% CI 0.96–1.17), and 1.12 (95% CI 1.01–1.23), respectively. A significantly positive association between frequent use of acupuncture and breast cancer risk implied certain unadjusted confounders that could exaggerate the risk of breast cancer among TCM users.

Access to the research data was approved by the National Health Research Institutes Review Committee (Approval number NHIRD # 100206).

Informed consent of the study participants was not required because the dataset used in this study consists of de-identified secondary data released for research purposes.

This population-based case-control study suggested that exposure to Angelica sinensis showed a weakly but significantly protective effect on breast cancer risk. The results remained intact after various falsification approaches and sensitivity analyses. Although limited by potentail souces of bias, our study tended to support the potentail protective effect from exposure to Angelica sinensis , which could ease people’s concern over the potential carcinogenic effect from exposure to Angelica sinensis .

This study found that the use of Angelica sinensis manifested a weak and protective effect to breast cancer risk after we adjusted for potential confounders. Furthermore, the dose-response relationship displayed the increasing protective effect in association with an increase in the doses of single and compound prescription of Angelica sinensis . We also observed a tendency that the earlier the initial use of Angelica sinensis happened, the stronger the seemingly protective effect of Angelica sinensis was observed. The findings mentioned above revealed that there was no evidence suggesting an increased breast cancer risk, but instead a small protective effect from the use of Angelica sinensis . Furthermore, the study findings also conformed to the current literature which revealed that TCM could provide a protection effect in breast cancer patients. 11 A recent literature review revealed that Angelica sinensis did not pose stimulatory effect on breast cancer in both in vitro and in vivo studies , which largely removes people’s fear of Angelica sinensis 12 . The previous study had observed that the extracts of Angelica sinensis , on contrary with previous cell culture, possessed the potential of anti-estrogen effect under the condition of the estradiol existing in cells. 5 In Taiwan, recent studies based on the NHIRD also provided support for the protection role of Angelica sinensis . For example, Siwutang (contain 25% Angelica sinensis ) and single Angelica sinensis were both associated with lower risk of and better outcome for breast cancer patients 13 , 14 . Epigenetic research also found that extraction of Angelica sinensis Z-ligustilide could restore the inhibitory effect of anti-hormone drug tamoxiphen on breast cancer cells, and suggested that it may be used as an adjuvant in the hormone therapy. 15 , 16 Furthermore, we found in the age-stratified analyses that the protective effect of Angelica sinensis was not statistically significant in the sample after menopause, which also suggested that the protective effect was correlated with the existence of estrogen. Future studies may be carried out to examine whether our study findings can be reproduced. Beside phytohormone mechanisms, other research revealed that the polysaccharide of Angelica sinensis could activate caspase-3 protease by cyclic AMP response element binding protein to facilitate the apoptosis of breast cancer cells. 6 In addition to the studies of other cancer, N-Butylidenephthalide, which extracted from Angelica sinensis , could induce p53 pathways contributing to the apoptosis and anti-proliferative effect in glioblastoma multiforme, liver and colon cancer cells. 17 , 18 , 19 And the polysaccharide APS-2a, extracted from Angelica sinensis , could also inhibit the proliferation of transplanted sarcoma. 20 Our study has a number of strengths. First, the current evidence on whether using Angelica sinensis could enhance the risk of breast cancer mostly comes from cell experiments. To the best of our knowledge, our study is s the first population-based cohort study that analyzed the risk of breast cancer in association with exposure to Angelica sinensis . Second, this study used a random sample of Taiwan’s NHIRD, which covers medical claims of more than 99.9% of Taiwanese residents. With such population-based medical claim data, the potential for selection bias was considered small. Both cases and controls were sampled from the same population, and such nested case-control design further provides reassurance that the potential for selection bias is minimal. Most importantly, the time density sampling method was used in this study, which increased the comparability between cases and controls with respect to the potential time-related confounding. There were several limitations involved in this study that should be addressed. First, self-paid medications were not included in the NHIRD, which could result in erroneous ascertainment of Angelica sinensis exposure. However, the potential bias resulting from such exposure misclassification is likely to be non-differential, which would tend to attenuate rather than overestimate the association of Angelica sinensis with breast cancer. Secondly, we can only control the potential confounders available from the NHIRD, and were unable to manage the potential confounding by some other known risk factors for breast cancer, especially those reproductive and genetic risk factors for breast cancer. Despite that, there were no apparent associations of prescription of Angelica sinensis with reproductive and genetic risk factors for breast cancer. In addition, we managed to control for the residential area and insurance premium based monthly salary, which may help reduce the potential confounding by several socioeconomic related risk factors for breast cancer, such as education, times of pregnancy and breastfeeding. Furthermore, we used acupuncture habit to verify the correction in the socioeconomic status of TCM users as falsification analysis, the results of which still supported the outcome of this research in high credibility. Third, in Taiwan, the same item of scientific TCM powder, made in different pharmaceutical companies (i.e, different brands), may follow the different dose reference, which makes it very difficult even not impossible to figure out the exact ratio of Angelica sinensis from each drug code. Thus, we deliberately used 10% to determine the Angelica sinensis dose for prescribed compounds with 10% or higher percentage of Angelica sinensis . A potential underestimation of Angelica sinensis exposure in this study would result in exposure misclassification. However, such exposure misclassification is likely to be non-differential, which again would lead to underestimate rather than overestimate the association between Angelica sinensis and breast cancer.

Since hormone replacement therapy had been found to increase the risk of breast cancer, 1 , 2 the scientists also had concerns about the breast cancer carcinogenic effect of phytohormone contained in herbs of traditional Chinese medicines (TCM), especially Angelica sinensis (Dang-guai), which is commonly used to improve gynecological disease even as flavoring in many Asia countries as well as countries outside of Asia where Angelica sinensis is also commonly used. In the related research of Angelica sinensis , some researchers found that Angelica sinensis extract could stimulate the proliferation of breast cancer cells. 3 , 4 However, the results observed in the subsequent experiments were not consistent. It was even observed that some Angelica sinensis extract had anti-cancer effects such as inhibiting estrogen activity or promoting apoptosis of breast cancer cells. 5 , 6 By far, there is no consensus on the issue of Angelica sinensis and breast cancer risk relationship. Among the top ten common used TCM herbs in Taiwan, two compound contain Angelica sinensis . 7 In addition, there were about 30% of patients in Taiwan use Chinese medicine after been diagnosed with breast cancer, 8 and the top ten compound drugs used in breast cancer patient have seven compounds contain Angelica sinensis . 9 To sum up, Angelica sinensis was very common prescribed in general public and breast cancer patients in Taiwan, but there was still no evidence of large epidemiological literature on the exact relationship between exposure of Angelica sinensis and breast cancer. We conducted this population-based epidemiological study to investigate whether intake of Angelica sinensis preparations by women was associated with breast cancer incidence.

The authors declare that they have no conflict of interest.