Invited Perspective: Closing the Loop of Nonmonotonicity—from Natural Hormones in Experimental Endocrinology to Endocrine Disruptors in Epidemiology

Most environmental epidemiological and regulatory studies assume that chemical exposure results in unidirectional (monotonic) dose–response curves, so that the higher the dose, the larger the effect. The alternative, a nonmonotonic dose–response curve (NMDRC), is defined as a nonlinear relationship between dose and effect that is characterized by a change of sign of the slope of the curve within the range of doses examined. An example of one type of NMDRC can be found in the work by Costas et al. in this issue of Environmental Health Perspectives, which revealed for the first time an inverted-U risk trend between total effective xenoestrogen burden (TEXB) levels in serum and endometrial cancer.¹ These findings involved levels of both the $\text{TEXB-}\mathrm{\alpha }$ (nonpolar) and $\text{TEXB-}\mathrm{\beta }$ (polar) fractions; the latter includes endogenous hormones such as estradiol and xenoestrogens such as bisphenol A (BPA).

To provide context for the importance of these findings, it is necessary to revisit some of the work that preceded this study by Costas et al. We have been fortunate to collaborate with two of the authors of this paper, Nicolás Olea and Marieta Fernández, on projects involving nonmonotonicity and mixtures of xenoestrogens. During Olea’s stay in our laboratory as a Fulbright Scholar, we jointly described the nonmonotonic response to androgens of a prostatic cancer cell line.^2,3 Later, during Fernández’s stay, we developed the total xenoestrogen burden assay.^4,5 Transatlantic collaborations continued for over a decade, including our sabbatical in Olea’s lab in 1996.

While Olea was working on the proliferative response to testosterone in our lab, we discovered the first xenoestrogen shedding from plastics—p-nonylphenol. Its presence in centrifuge tubes occurred due to a formulation change by the manufacturer.⁶ At that time, various studies were exploring correlations between exposures to single xenoestrogens, such as the insecticide DDT, and breast cancer incidence. However, given the increasing list of environmental estrogens—of which $p$-nonylphenol was only the latest identified—we decided to measure the effects of combined exposure to xenoestrogens rather than relate an outcome to a single chemical. This approach inspired us to devise a method of measuring the total xenoestrogen burden.^4,5 While our group continued to work on theoretical and experimental biology using cell cultures and animal models, Olea and Fernández returned to Spain, where they optimized the total xenoestrogen burden methodology and applied it to biomonitoring of environmental exposure in human tissues⁷ and in epidemiological studies of breast cancer,⁸ hypospadias,⁹ cryptorchidism,⁹ and now endometrial cancer.¹ Meanwhile, Kortenkamp’s group published the first of a long series of articles addressing mixture effects from combinations of xenoestrogens, each at a level that produced no effect alone.¹⁰

Why is the nonmonotonicity of hormones (and endocrine disruptors) relevant to endocrinology, oncology, and therapeutics? By the time we published on the nonmonotonicity of androgens, it was already known that ovarian estrogens are an important risk factor for breast cancer.^11–14 Experimental carcinogenesis in animal models also revealed that physiological estrogen levels were necessary for the development of mammary cancer after exposure to chemical carcinogens.^15,16 In contrast, supraphysiological doses of estrogens increased the latency period of chemically induced mammary tumors and reduced the number of tumors induced by these agents.^17,18 These findings suggested that high doses of estrogen could be of therapeutic use for breast cancer; in fact, diethylstilbestrol treatment was effectively used for this purpose until the advent of antiestrogens, which were equally effective but produced fewer undesirable side effects.^19–21 Endocrinologists studying the processes that underlie these opposite effects found a time-dependent inhibitory effect of a second estrogen dose on the magnitude of the first response to estradiol on uterine DNA synthesis, a phenomenon Gorski called “shutoff.”²² In addition, Amara and Dannies found a dose-dependent effect of estrogens on rat pituitary cell lines, revealing a biphasic effect on cell number.²³

Our work exposing androgen-responsive cell lines to different concentrations of androgens also revealed an inverted-U curve, whereby low concentrations increased the cell number and high concentrations decreased it. In addition, during the initial proliferative phase there was very little induction of prostate-specific antigen (PSA), whereas during the later proliferative inhibitory response PSA reached a maximal induction that plateaued. Using somatic cell genetics methodology, we were able to separate the ascending from the descending part of the proliferative curve in discrete cell lines; namely, each cell line expressed only one of the two distinct phenomena in a monotonic pattern.^3,24 Although endocrinologists had no problem accepting these nonmonotonic dose–response curves, which are more the norm than the exception in endocrinology, most toxicologists working for regulatory agencies and industry resisted accepting this pattern.²⁵

Although NMDRC have been reported in epidemiological surveys,²⁵ Costas et al. are now revealing for the first time that both $\text{TEXB-}\mathrm{\alpha }$ and $\text{TEXB-}\mathrm{\beta }$ measurements of mixtures of estrogenic compounds relate to endometrial cancer in a nonmonotonic dose–response curve relationship. The half-century-old evidence obtained from the carcinogenesis experiments mentioned above, plus the equally old evidence of the effective therapeutic use of estrogens to induce cancer regression, illustrate it is high time for epidemiologists to stop expecting monotonic dose–response curves as the norm. Moreover, the extensive documentation of nonmonotonic dose–response curves in endocrinology in general and more recently for endocrine disruptors,^26–30 is additional compelling evidence for this proposed paradigmatic change in epidemiology.

In the span of more than 3 decades since we and others pioneered the field of endocrine disruptors,³¹ overwhelming evidence has been gathered in animal models and epidemiological studies showing that exposure levels to certain endocrine disruptors, such as BPA, are above those that produce deleterious health effects in animals.³² It is now well established that humans are exposed to mixtures of numerous endocrine-disrupting chemicals and that these mixtures can produce adverse health effects.³³ In this regard, in 2023 the European Food Safety Authority significantly lowered the tolerable daily intake (TDI) of BPA and concluded that “consumers with both average and high exposure to BPA in all age groups exceeded the new TDI, indicating health concerns.”³⁴

Finally, when interpreting data, researchers should acknowledge the meaningful results generated by “descriptive” studies; they provide a firm basis on which to build a solid theoretical framework. When we assess the findings gathered in different areas of endocrinology together with the present results, we now observe a full and coherent cycle.

Conclusions and opinions are those of the individual authors and do not necessarily reflect the policies or views of EHP Publishing or the National Institute of Environmental Health Sciences.

Refers to https://doi.org/10.1289/EHP13202