The Use of Herbal Preparations to Alleviate Climacteric Disorders and Risk of Postmenopausal Breast Cancer in a German Case-Control Study

Over the past several years, awareness has risen that the use of menopausal hormone therapy (HT) is a risk factor for breast cancer. As a result, increasing numbers of women are turning to herbal preparations (HEP) to manage their menopausal symptoms. A wide range of HEP are offered for that purpose, including phytoestrogens (flavonoids, stilbenes) and other mixed polyphenolic substances with less specified pharmaceutical agents from plant species [Actaea racemosa (called Cimicifuga throughout the text), St. John's Wort, Vitex agnus castus, Pulsatilla pratensis, Rheum rhaponticum, Ginseng etc.). Many of these bind competitively to the estrogen receptors (ER) α and β and exert low estrogenic and/or antiestrogenic activity (1, 2). Furthermore, some have been shown to interfere with enzymes responsible for biosynthesis and metabolism of estrogens. Aside from ER-mediated activity, these substances can suppress cell cycle involving growth factors and cell signaling pathways, and promote hormonally independent antioxidative actions and apoptosis (3-5).

It has been postulated that isoflavones protect from breast cancer (6, 7), but epidemiologic studies on dietary intake of soy isoflavones have found little evidence for the potential to prevent breast cancer in postmenopausal women from western countries, where consumption is low (<1 mg/day; refs. 8, 9). Daily intake of HEP ingredient supplies a much higher amount of isoflavones (25 to 300 mg/day) than that found in the average western diet.

Since findings from toxicologic studies on flavonoids have observed cell growth of breast cancer cell lines in animal models at low physiologic concentrations (5, 10-12), concerns have risen about the potential risks of long-term use of phytoestrogen preparations, especially in women high at risk of breast cancer (13, 14).

Antiproliferative, antiestrogenic properties of the root extract of black cohosh (Cimicifuga) on receptor-positive and receptor-negative breast cancer cell lines in vitro have been reported by some studies, whereas in others, no such effects were seen (2, 11, 15). In a recent clinical study, an isopropanolic extract of black cohosh (Remifemin) had no effect on breast density or cell proliferation in postmenopausal women (16), whereas observational studies have shown protective effects on occurrence and recurrence of invasive breast cancer (17, 18).

No study has exhaustively evaluated the extent to which a healthy life-style may confound the effects of HEP use. Apart from higher education and lower age, regular exercise was observed to be related to the use of menopausal herbal medications (19-21).

Our objective was to analyze associations between patterns of use of HEP and invasive breast cancer in postmenopausal women, thereby investigating whether life-style factors such as a healthy diet, physical activity, and previous HT may act as confounders. Second, we analyzed differential risks of HEP by tumor histology and receptor status, to inform discussions on the potential modes of action of HEP.

The Mamma carcinoma Risk factor Investigation (MARIE) study was a population-based case-control study conducted between August 2002 and September 2005 in two regions in Germany, the Rhine-Neckar-Karlsruhe region (RNK) and the city of Hamburg. The study was approved by ethics committees and conducted in accordance with the Declaration of Helsinki. All participants gave written informed consent. Details of the study protocol have been published elsewhere (22). In short, postmenopausal patients ages 50 to 74 y were identified through participating clinics and the Hamburg cancer registry, and deemed eligible if they had a histologically confirmed first primary invasive or in situ breast cancer. Of 6,114 eligible patients, 3,919 (64.1%) participated. For each participating case, two randomly selected registry-based controls were frequency matched on study region and year of birth. Of 17,093 eligible controls, 7,421 (43.4%) participated. Of the control women who refused participation, 2,966 answered a short questionnaire by telephone, which covered key variables of the MARIE questionnaire. In general, we had no access to contact information of nonparticipating cases due to data protection regulations. As an exception, the short questionnaire was offered to 317 cases from one clinic in Hamburg, of which 144 answered.

Information on risk factors for breast cancer was collected in face-to-face interviews, focusing on hormonal (classified into monoestrogen, combined estrogen and progesterone, tibolone, and unknown HT; see ref, 22 for details) and herbal therapies for menopausal disorders. Each form of HEP (tablets, injections, teas, and homeopathic pills) and duration of use was recorded, supported by an alphabetical list of herbal products, brand names, and respective year of release to the German market. Constituents of most specified preparations were obtained from a list of pharmaceuticals in Germany (23) and by search of the internet. Upon screening for ingredients, HEP were classified as one of seven main groups, (a) Remifemin (Cimicifuga) or Remifemin plus (Cimicifuga, St. John's Wort), (b) other Cimicifuga products, (c) St. John's Wort, (d) Vitex agnus castus, (e) “phytoestrogens” (soy isoflavones, red clover), (d) otherwise specified (mainly Pulsatilla, Rheum rhaponticum), and (d) unknown preparations. Women who used mixed or different preparations were classified in more than one HEP class.

Physical activity and dietary habits were considered as potential confounders for the effect of HEP. Various physical activity variables were available and tested in preliminary models. Exercise was more closely related to HEP use than leisure time physical activity, and thus, the former was finally selected. Exercise was assessed by asking participants to specify up to three types of sports done during two age periods (30-49, 50+ y), including duration and frequency. A metabolic equivalent (MET) value was assigned to each reported activity according to the classification of Ainsworth et al. (ref. 24; e.g., jogging, 7 MET; calisthenics, 4.5 MET). MET-hours per week were calculated by summing the average hours per week spent on sports weighted by the MET values over the whole period since age 30 y (25). Finally, based on the distribution in controls, quintiles of MET-hours per week were used for analysis.

Dietary habits during the past 12 mo were collected from 5,616 (84.5%) controls and 2,942 (85.0%) cases with a self-administered validated 176 items' food frequency questionnaire adapted from the German centers of the European Prospective Investigation into Cancer and Nutrition (26). Nutrients were calculated using the “German Nutrient Data Base BLS II.3” (Federal Research Centre for Nutrition and Food). Dietary habits included daily intake of fruits, vegetables, herbal tea, black/green tea, coffee, total fiber, and soy.

To adjust for age differences between cases at the date of diagnosis and controls at the date of interview, a reference date to the date of diagnosis was created for controls by subtracting the center- and birth year–specific mean age difference between cases and controls in days from the date of interview of controls (22).

Women were considered to be “ever” users of HEP if they reported use for >3 mo before the reference date in controls or the date of diagnosis in cases. Ever users of HEP were classified as “current” users if ever-use occurred within 6 mo before the reference date; otherwise, ever users were considered “past” users. The same definitions were applied to the use of HT.

Unconditional logistic regression analysis generating adjusted odds ratios (OR) with 95% confidence intervals (95% CI) were done. For the analyses by tumor characteristics (histologic type: invasive ductal, invasive lobular, mixed ductal/lobular or tubular combined, carcinoma in situ; ER, progesterone (PR), and her2neu receptor status (a prognostic marker of human epidermal growth factors) polytomous logistic regression models (SPSS version 15) were applied. Different exposure variables were used to test for associations with the same reference “never use of HEP” in all models [ever use; past or current use, duration of use in categories (<1; 1-5; >5 y) and as a continuous variable, and seven classes of HEP ingredients simultaneously].

Statistical power yielded 84% for the total study group (with an exposure prevalence of 10.1% as in controls and an assumed OR of 0.80 at α = 0.05), whereas in small subgroup analyses (i.e., Her2neu-positive tumors), it was as low as 54% to detect a risk reduction by 30%.

Models were adjusted for a set of covariates that included known risk factors or statistically significant covariates in multivariate models, and the matching variables (for categories, see Table 1 and footnote of Table 2): study center, year of birth, use of HT, age at menopause, age at menarche, number of full-term pregnancies, ever breastfeeding, number of mammograms, ever benign breast disease, first-degree family history of breast cancer, occupational status, and exercise. Diet variables (categorized in quartiles based on the distribution among controls) were simultaneously entered into the final model, which was adjusted for total energy intake.

Effect modification was assessed by testing for multiplicative interaction using a cross-product term for the variable of interest (HEP) and potential interaction variables in the multivariate-adjusted model. Potential interaction was tested separately for exercise, HT, and an index for healthy diet. Significance of the interaction term was evaluated with the likelihood-ratio test. Pair wise differences between effect estimates of HEP use among tumor characteristics were tested for heterogeneity, taking the covariance of the effect estimates into account (27). For heterogeneity testing of estimates of HEP classes from a simultaneous model of all invasive breast cancer, the weighted sum of squared differences between individual HEP class effect and the pooled effect across seven HEP classes was calculated, with the weights being the inverse variance of the estimates. The test value is distributed as a χ² statistic with 6 degrees of freedom.

Recently published results from the MARIE study suggested that cessation of HT is accompanied by a leveling of the higher breast cancer risk observed in current HT users (22). Therefore, any association between HEP and breast cancer might be confounded by HT use. To meet these concerns, in addition to adjusting for HT, a sensitivity analysis was conducted by restricting the study group to women who had never used HT. In this subset, the statistical power to detect a significant OR of ∼0.70 for ever HEP use was 78% at α = 0.05.

The characteristics of the 10,121 postmenopausal women (3,464 cases, 6,657 controls) included in this analysis have been described elsewhere (22). Briefly, 55% of the women were recruited in Hamburg and 45% in the RNK region. Mean age was 63.2 years in controls and 63.3 years in cases. Women, who had ever used HEP (9.9%), were younger than nonuser of HEP, less likely to have an unknown age at menopause, more likely to have reported benign breast disease, had a higher occupational status, and a lower body mass index (BMI; Table 1). Furthermore, HEP users were more frequently low alcohol consumers, past user of HT, physically more active, and more frequently consumed a diet rich in tea compared with nonusers of HEP. These differences were similarly found in cases and in controls (Table 1).

In the subgroup of women who had never used HT (n = 1,059 invasive cases, 2,683 controls), a higher proportion lived in the RNK region, reported a natural menopause, had fewer mammograms, lower occupational status, had a BMI over 25 kg/m², and exercised less frequently than the total group (data not shown).

Ever use of any HEP was reported by 224 (6.7%) invasive cancer cases and 669 (10.1%) controls (Table 2). More women were past users of HEP compared with current users (5.6% versus 3.3%). Current HT users were less likely to use any HEP (cases 4.5%, controls 5.4%).

Compared with never users, ever (OR, 0.73; 95% CI, 0.62-0.86), past (OR, 0.75; 95% CI, 0.62-0.92), and current use of HEP (OR, 0.70; 95% CI, 0.53-0.92; Table 2) showed negative associations with invasive carcinomas that were statistically significant. ORs of past versus current use did not differ significantly (P_{heterogeneity} = 0.66). Differences in risk by categories of duration of HEP use were small, but increasing duration of HEP use was associated with an inverse trend for invasive breast cancer (P_trend = 0.03, Table 2). There was no significant effect modification for past/current use of HEP with past/current use of HT (P_interaction = 0.95) or sport (P_interaction = 0.22). Risk estimates for HEP use were not affected by adjustment for any of the dietary variables (total fiber, soy intake, vegetables, fruits, coffee, black/green tea, and herbal tea).

In the sensitivity analysis restricted to women who had never used HT, ever use of any HEP was consistently associated with a reduced risk for invasive breast cancer (OR, 0.72; 95% CI, 0.56-0.95), without variation by past, current, or duration of HEP use (P_trend = 0.18, Table 2). Again, associations were not modified by sports for using a binary (ever HEP, P_interaction = 0.36) and a categorical variable for HEP (past/current HEP, P_interaction = 0.17).

Among HEP classes of active ingredients, Cimicifuga was the most frequently used HEP, primarily as Remifemin/Remifemin plus. Overlapping of HEP classes occurred in 9% of mentioned HEP mainly between phytoestrogens and Remifemin (n = 30) and between Vitex agnus castus and Remifemin (n = 7), other Cimicifuga (n = 10), and “otherwise specified” HEP (n = 7). There was no significant heterogeneity of risk estimates of the seven classes of HEP (P_{heterogeneity} = 0.81). Inverse associations with invasive breast cancer were found for Remifemin/Remifemin plus (OR, 0.80; 95% CI, 0.63-1.003), Vitex agnus castus (OR, 0.40; 95% CI, 0.17-0.97), otherwise specified preparations (OR, 0.48; 95% CI, 0.25-0.93), “phytoestrogens” (OR, 0.64; 95% CI, 0.39-1.05), and “unknown” preparations (OR, 0.77; 95% CI, 0.54-1.08; Table 2).

Considering receptor status, ever use of any HEP was inversely associated with both positive and negative ER (OR_ER⁺, 0.74; OR_ER⁻, 0.68) and PR status (OR_PR⁺, 0.77; OR_PR⁻, 0.66). Risk estimates by ER and PR status proved to be homogeneous (ER⁻ versus ER⁺ P_{heterogeneity} = 0.64; PR⁻ versus PR⁺ P_{heterogeneity} = 0.29; Table 3). Duration of HEP use was significantly inversely related only to PR⁻ tumors (P_trend = 0.01; Table 3).

Regarding her2neu status (Table 3), estimates of ever use of HEP were similar for tumors with negative (OR, 0.75) and positive status (OR, 0.73), whereas the duration of HEP use was not associated with Her2neu status.

For invasive ductal cancers, risk estimates of ever, past, and current HEP use were similar to those for all invasive cancers (ever HEP, OR 0.72; Table 3). The same was true for combined lobular, mixed ductal/lobular, and tubular tumors (ever HEP OR, 0.76; Table 3) with past and current use of HEP. Duration of HEP use was significantly inversely related to risk of lobular/mixed/tubular (OR, 0.92 per year; P_trend = 0.03) but not to ductal tumors. HEP use was not related to in situ carcinomas (Table 3).

This study provides evidence that use of HEP to manage climacteric disorders may have a protective effect on invasive breast cancer in postmenopausal women, independent of other health behaviors. The risk for invasive breast cancer was reduced by ∼26% for ever use of any HEP and by 4% per year of use. Healthy life-style, represented by physical activity and dietary variables, did not affect this relationship. The HEP-associated risk was not substantially modified by histologic type and receptor status of the tumor. Risk estimates did not differ significantly between HEP classes of active ingredients.

Participation bias could have resulted in an overrepresentation of controls using HEP, which would have led to an overestimation of the inverse association with breast cancer. To assess the extent of such bias, we asked nonresponding women in a short questionnaire whether they had ever used any HEP. Among the 1,820 short questionnaire respondents from Hamburg, 89 (5.3%) controls and only 3 cases (2.1%) reported ever having used HEP. Because of the higher proportions of HEP use among both cases and controls included in the study, we conclude that compared with the population, controls and possibly cases were more likely to have overreported than to have underreported HEP use. Some of the differences in HEP use between study group and short questionnaire participants might be explained by social and demographic differences, as nonparticipants were older and less well-educated (19, 20, 22). As the relative difference of HEP use between cases and controls pertained in nonresponders, it is unlikely that the direction of the association is explained by a nonresponse bias.

The MARIE questionnaire has been evaluated and shown to be of good reliability according to HT use. Furthermore, women's self-reporting of HT use was in good agreement with physician reports on HT prescriptions (28, 29). HT use was reported by 63.9%, whereas only 9.9% reported having used HEP, but within groups of invasive cancer cases and controls, HT users were equally likely to have used HEP as nonusers of HT (cases 6.5% versus 7.7%, Pχ² = 0.21; controls 9.7% versus 10.7%, Pχ² = 0.20). Thus, it is unlikely that differential recall of HEP use by HT status would have affected our results.

Misclassification could have occurred due to difficulties in recalling brand names (30). To disentangle active ingredients, we analyzed HEP classes simultaneously in a single model, thereby controlling for each. However, results of single HEP classes were based on small numbers. Moreover, information on individual dosages was not available. Although most of these HEP are well-regulated and standardized pharmaceutical products in Germany (31), producer recommended dosages for a specific HEP vary according to severity of complaints.

To date, one observational study has investigated the invasive breast cancer risk with respect to menopausal HEP use (17). Concordant with our finding, a preventive effect of ever use of any HEP was observed in that study (OR, 0.65; 95% CI, 0.49-0.87), which was slightly larger than the effect in our study.

Relative risk estimates in our study were adjusted for potential confounders, including current and past use of HT, physical activity, and diet. BMI, food variables, alcohol, and smoking were not significantly associated with invasive breast cancer and did not confound associations of HEP, in contrast to HT (22) and physical activity (25). None of the life-style factors significantly modified the risk of HEP use on invasive breast cancer. This was supported by our sensitivity analysis of HEP use in women, who had never used HT, which showed the same inverse association as the unrestricted analysis. Yet some residual confounding by unmeasured variables or imprecise information is possible. Dietary and physical activity data were not accounted for in the previously published U.S. study (17).

Among active ingredients of HEP, soy isoflavones, flavonoids, and stilbene derivates were reported to have antiestrogenic as well as estrogenic properties in breast tissue through their affinity to ER. In our study, a nonsignificant OR for HEP comprising isoflavonoids from soy and red clover (called phytoestrogens) compared well to that reported by Rebbeck et al. (17). A recent meta-analysis on dietary soy intake and breast cancer found that soy isoflavones may protect from breast cancer when consumed at high amounts (9).

Two other classes of HEP in the present study, Vitex agnus castus, containing flavanols, and other specified HEP (mainly Pulsatilla, Rheum rhaponticum), containing stilbenes, were shown for the first time to potentially protect from invasive breast cancer.

Cimicifuga (black cohosh) is not considered to bind to ER and to exclusively exert antiestrogenic effects (11). We observed Remifemin (Cimicifuga)/Remifemin plus (Cimicifuga, St. John's Wort) to be weakly inversely related to invasive breast cancer, whereas other Cimicifuga products were not. In contrast, Rebbeck et al. (17) found overall Cimicifuga products to be associated with breast cancer. In a retrospective cohort of invasive breast cancer patients, Remifemin was shown to prolong the time to relapse (18).

We have analyzed effects of the Cimicifuga classes “Remifemin/Remifemin plus” and “other Cimicfuga products” on ER and PR status to compare our data with published data from Rebbeck and colleagues (17), who observed protective effects of Cimicifuga products in ER⁺, ER⁻, and PR⁺ tumors, whereas numbers for PR⁻ tumors were too small to interpret. In our analysis, Remifemin/Remifemin plus was inversely associated with breast cancer risk, independent of ER (ER⁺ OR, 0.73; 95% CI, 0.56-0.94; ER⁻ OR, 0.80; 95% CI, 0.54-1.20) and PR status (PR⁺, 0.81; 95% CI, 0.62-1.05; PR⁻ OR, 0.69; 95% CI, 0.48-0.98). Thus, our results for any HEP and for Remifemin do not support substantial differences in associations with positive or negative ER/PR status of the tumor. No further study has addressed the relationship between HEP use and breast tumor characteristics.

It is noteworthy that nearly 50% of ERα-negative tumors were estimated to express ERβ (32, 33), which was not assessed in our study. Phytoestrogens preferably bind to ERβ (34, 35), and activated ERβ may reduce cell proliferation independently of ERα (36), a mechanism that may have been operative in ER⁻ and PR⁻ cases. Competitive binding of HEP agents to ER or PR may also have contributed to cell growth inhibition in receptor-positive tumors because estradiol levels are low in postmenopausal women. In line with our results are possible pathways of HEP acting independently of ER or PR. HEP ingredients have been shown to interfere with enzymes of estrogen biosynthesis by transversion of estrone to estradiol and androstenedion to testosterone and vice versa (1). At high breast tissue concentrations, HEP ingredients may induce cell cycle arrest and apoptosis (4, 37-40) or may function as antioxidants by protecting cells from DNA damage (36, 41). Those mechanisms that work independent of ERα and PR seem to contribute to the observed effects of HEP.

In summary, we conclude that in postmenopausal women HEP use may exert a protective effect on risk for invasive breast cancer, irrespective of histologic type and receptor status. The specific ingredients responsible for this potential benefit need to be further elucidated.

No potential conflicts of interest were disclosed.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank all participants and clinical colleagues for supporting the study and “Deutsche Krebshilfe e.V.” for funding the MARIE-study (grant 70-2892-BR I).