Hormone‐related supplements and breast cancer risk: Need for improved measurement of supplement use

Alternative supplements such as herbal and other non-vitamin, non-mineral compounds may contain substantial amounts of bioactive ingredients, but despite widespread use, little is known about long-term risks or benefits.1 Rebbeck et al. are to be congratulated for presenting data that address this gap, examining hormone-related supplements and breast cancer risk, for the first time.2 We suggest, however, that aspects of their exposure measurement and results need to be reconsidered, to advance future risk assessment. We wish to primarily comment on three measurement limitations: the exclusion of supplements which are frequently used; the exclusion of supplements containing considerable amounts of hormone-related components; and the inclusion of items and decisions that encouraged misclassification. Since estrogen-like components such as phytoestrogens may be associated with reduced breast cancer risk,3, 4 amending Rebbeck et al.'s underestimates of hormone-related supplement intake would reduce exposure misclassification, and observed associations with cancer risk would be more accurate. First, despite difficulties comparing supplement use due to varying inclusions, ever use should equal or surpass current use when items and populations are similar. However, among similarly aged Caucasian and African American women, Rebbeck et al.'s ever use estimate is lower than those reporting current use.5-7 This may be partly explained by Rebbeck et al.'s short, specific list of hormone-related items, since estimates of use depend on which and how many supplements are queried. However, their estimates are also low compared with those reporting supplement use for menopausal symptoms,8, 9 suggesting Rebbeck et al. excluded certain hormone-related items. Recent work may shed light on these exclusions. We reported that 31% of women (primarily Caucasian) ever used hormone-related supplements,9 considerably higher than that reported by Rebbeck et al. for European American controls (17%). Our estimates were somewhat similar to Rebbeck et al.'s for five supplements (isoflavone, red clover, black cohosh, dong quai, ginseng), but a substantial proportion of women also reported using evening primrose oil.8, 9 This item and possibly other hormone-related supplements10 were omitted from Rebbeck et al.'s questionnaire. Second, supplements contain a variety of bioactive components and their contents are largely unknown,1 although some have been identified. In terms of hormone-related items, we and others have reported the phytoestrogen content of alternative supplements,11, 12 a class of substances mentioned by Rebbeck et al. Although they included two important phytoestrogen sources – soy and red clover (from which isoflavones, genistein, and daidzein may be derived) – other major contributors such as licorice and kudzu are absent.11, 12 Thus, omission of two informative parameters of intake (items either consumed frequently, or containing substantial amounts of hormone-related components) limits the completeness of Rebbeck et al.'s assessment, but also suggests items to be considered in future studies. A third and final limitation is that of misclassification due to questionnaire items and analytic decisions that encouraged inaccurate reports. Survey respondents may be confused about how herbal or alternative supplements differ from traditional vitamin or mineral supplements, and the contents of what is consumed.13, 14 This was illustrated in Rebbeck et al.'s description of women reporting Remifemin (whose main ingredient is black cohosh10), but not black cohosh. Given this confusion, it is important to clearly define and query supplement use, particularly when researchers and respondents perceive these items differently.6, 14 Although it is not clear how Rebbeck et al. specifically asked about supplement use, it may have been misleading to include two ‘natural’ hormone replacement medications, Biestrogen and Triestrogen, as hormone-related supplements. Natural hormone replacement medications are only available by prescription,15 making them legislatively and perceptually different from dietary supplements.16 This may have contributed to their extremely low reported use, when they have otherwise been described as popular hormone replacement therapies.17 Furthermore, it is not clear how their inclusion as supplements affected the reported ‘use of exogenous hormones and use of other medications,’ or identification as a confounder (as ‘ever use of hormone replacement therapy’). A similar issue concerns the inclusion of steroid and yam creams, since these topical items are not intended for ingestion, a hallmark characteristic of ‘supplements’.1, 16 Also problematic is the listing of Promensil and red clover as two independent items, when Promensil is a major brand name supplement used in red clover clinical trials.10 Rebbeck et al. also unwittingly restricted data collection by limiting responses to a maximum of five supplements. Although few women report current use of five or more supplements,18 ever use introduces an expansive repertoire of items that could easily surpass five, especially given the overlap of red clover items just mentioned with derivative compounds that were listed separately: isoflavones, genistein and daidzein. Other overlapping items, and brand names containing listed supplements as ingredients compound this problem (black cohosh and Remifemin; soy medications, isoflavones, genistein and daidzein; Rejuvex containing dong quai; Estrovin containing soy and black cohosh). Thus, of 17 supplements listed on their questionnaire, 11 overlap with at least one other item, and four are not supplements in the traditional sense, being either unavailable except by prescription or not intended for ingestion—a context likely to promote uncertainty and inaccurate response. Future work must be attentive to these and other pitfalls of intake measurement, and the inclusion of clearly identified items to encourage accurate response and subsequent validation of measures will be of great importance in this new frontier of alternative supplement research.1 Although mainly concerned with the measurement issues just discussed, we also wish to draw attention to possible errors in Rebbeck et al.'s results. Although 280 controls and 123 cases are reported as supplement users in Table III, these numbers are surpassed or misreported in Table II for age at menarche and parity among users. Since it is uncertain which numbers are correct or how errors affect results, we encourage the authors to reexamine their data and report corrected findings. We also wish to comment on certain findings, although these may shift after data are corrected. Menopausal status was not found in Table II to be a risk factor for breast cancer, although reported as such in the text. Instead, oral contraceptives, as well as family history and parity were associated with risk among non-users of supplements. In contrast, most reproductive factors were not associated with risk among supplement users. This intriguing finding may warrant further discussion if it persists after data correction. In conclusion, Rebbeck et al. have presented interesting work that contributes needed data on the association between hormone-related supplements and cancer risk. Given the potential for widespread use as alternatives to conventional hormone therapy,19 and for contributing large doses of bioactive substances,11 hormone-related supplements merit additional research focus. Alongside this is the need to improve methods for collecting accurate intake data and to identify factors related to misreporting, and we offer our comments in the spirit of contributing to this need. Yours sincerely, Beatrice A. Boucher, Lilian U. Thompson, Nancy Kreiger, Michelle Cotterchio.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.