Genotoxic and genoprotective effects of phytoestrogens: a systematic review

Abstract Phytoestrogens are xenoestrogens found in plants with a myriad of health benefits. However, various studies reported the genotoxic effects of these substances. Thus, we reviewed in vitro and in vivo studies published in PubMed, Scopus, and Web of Science to evaluate the genotoxic and the genoprotective potential of phytoestrogens. Only studies written in English and intended to study commercially available phytoestrogens were included. The screening was performed manually. Moreover, the underlying mechanism of action of phytoestrogens was described. Around half of those studies (43%) reported genoprotective results. However, several studies revealed positive results for genotoxicity with specific model organisms and with dose/concentration dependence. The assessment of the selected articles showed substantial differences in the used concentrations and a biphasic response was recorded in some phytoestrogens. As far as we know, this is the first study to assess the genotoxic and genoprotective effects of phytoestrogens systematically.


Introduction
Phytoestrogens are natural xenoestrogens in the form of phenolic compounds, classified as plant-derived estrogens owing to their ability to mimic the action of estrogens (Zaheer and Humayoun Akhtar 2017). In plants, phytoestrogens are secondary metabolites, with no nutritional value but can exhibit a wide variety of biological activities to protect plants from biotic and abiotic stress (Tidke et al. 2017, Serra et al. 2018. Based on the number of phenolic rings, the type of substituents linked to these rings, and the structural elements bound to the rings and each other, these polyphenols can be divided into two main classes: flavonoids and non-flavonoids (Souto et al. 2019; Figure 1).
Most phytoestrogens found in the diet are isoflavones and lignans, presented mainly in soybean and its derivatives, legumes, vegetables, fruits, and cereals (Viggiani et al. 2019). In this respect, the average daily intake of isoflavones in Mediterranean countries (Italy, Spain, Greece, and Southern France) was 0.46 mg/day whereas in China and Japan the intake was 36.2 and 20.8 mg/day respectively (Hirose et al. 2005, S.-A. Lee et al. 2007, Zamora-Ros et al. 2012. However, a new study in Southern Italy reported a daily intake of 4 and 2.8 mg/day of isoflavones and lignans respectively (Godos et al. 2018). Notably, phytoestrogen exposure increased with the internationalization of soybean diets, the progressive popularity of legume-derived diets, and the use of phytoestrogen as supplements (Bennetau-Pelissero 2016, Isidoro et al. 2016). It seems that the global market of phytoestrogen supplements is anticipated to reach $4249.6 million by 2025, from $3540.4 million in 2019 (Market study reports 2020). Also, with the onset of the COVID-19 pandemic, isoflavones-based products emerged from 19 to 47 labels just in 2020 (US Department of Health and Human Services 2022). Yet, the required intake of phytoestrogens to improve vasomotor symptoms (an average of 50 mg/day) is higher than the daily intake (Desmawati and Sulastri 2019). Moreover, phytoestrogens bioavailability depends on multiple factors such as ethnicity, sex, hormone levels, health status, concentration, and phytoestrogen types (Dom ınguez-L opez et al. 2020).
Recently, several studies reported positive outcomes of phytoestrogens consumption on human health. For instance, meta-analyses indicated a significant decrease in the frequency of hot flushes of postmenopausal women after phytoestrogens consumption (Chen et al. 2015). Evidence suggests that phytoestrogens can also improve cognitive function (Cui et al. 2020), reduce the risk of obesity (Sandoval et al. 2020), and protect from diabetes mellitus (Loureiro andMartel 2019, Sun andMiao 2020). In addition, they possess anticancer properties (Torrens-Mas and Roca 2020, Cayetano-Salazar et al. 2021) by promoting apoptosis, inhibiting cell proliferation, and enhancing the immune system (Noriega-Rodr ıguez et al. 2020, Rizeq et al. 2020, S anchez-Valdeol ıvar et al. 2020. Further, isoflavones were effective against oxidative stress (Morvaridzadeh et al. 2020).
However, a high intake of soy-based foods during infancy may increase the risk of uterine fibroids in women (Qin et al. 2019) and induce heavy menstrual bleeding in adulthood (Upson et al. 2019). Moreover, neurobehavioral changes in mice were observed after recording gene expression changes in the hippocampus and hypothalamus and socio-communication impairments after treatment with genistein (Butler et al. 2020). In addition, after treatment with genistein, alterations of the dopaminergic, nitrergic, and vasopressinergic systems were observed in mice (Ponti et al. 2017(Ponti et al. , 2019. These findings thus challenge the concept of phytoestrogens as being solely beneficial compounds.
The term genotoxicity includes mutagenicity (i.e., irreversible and transmissible DNA damages) and also reversible and non-transmissible DNA damages that could be repaired through DNA repair processes or cellular processes (OECD 2016). In consequence, mutagenesis and genomic alterations may lead to several pathological endpoints including cancer (Barnes et al. 2018, Alhmoud et al. 2020. Therefore, to assure the safety of phytoestrogens, evaluating their genotoxic risk is recommended. Genoprotective effect refers to the effect induced by any agents that are involved in the protection against DNA damage through detoxification processes, DNA repair processes, and antioxidant activity (Koklesova et al. 2020). In fact, numerous studies have been conducted to evaluate the genotoxic and genoprotective effects of phytoestrogens. However, the results are still controversial. At which point a specific phytoestrogen is genotoxic or genoprotective is unknown yet. By assessing the genotoxic/genoprotective effect of phytoestrogens, we may help drug development based on natural products by identifying the stimulatory doses in each model organism and the mechanism of action to enhance the protective action of these substances. In this way, patients and healthy individuals will benefit from the genoprotective effect of phytoestrogens.
As a sequence, this review aims to give a better perspective on understanding the role of dietary phytoestrogens in human health, by deciphering their genotoxic/genoprotective effects through a systematic review in compliance with the PRISMA (Preferred Reporting Items for Systematic and Meta-Analysis) checklist. Further, the current review presented the underlying mechanisms of action of phytoestrogens.

Methods
To determine study eligibility, the population, intervention, comparison, outcomes, and study design eligibility criteria were used (Table 1). This review compiles the available data on phytoestrogens published from inception to December 2020 ( Figure 2). The search was carried out from PubMed, Scopus, and Web of Science databases. The following search words were used: ('phytoestrogen' OR 'dietary phytochemical' OR 'dietary phytoestrogen') AND ('mutation' OR 'mutagenicity' OR 'genotoxicity' OR 'DNA damage' OR 'antigenotoxicity' OR 'genoprotective'). Only those written in English and intended to study commercially available phytoestrogens (not originating from herbal extract) were included in the review to avoid heterogeneity yielding from the extraction process. The screening was performed manually. No registration number for the present review protocol. The retrieved articles were summarized in Supplementary  Table 1.

Primary outcomes
The authors considered mutations, micronucleus formation, DNA breaks, chromosome loss or gain, chromatid exchanges, and DNA repair gene expression changes as primary outcomes. These endpoints were assessed by various tests such as micronucleus assay, comet assay, and Ames test. All assays were presented in Supplementary Table 1.

Secondary outcomes
The authors considered changes in the level of the antioxidant enzymes, proteins, and oxidative stress biomarkers as secondary outcomes.

Time trends in publications
A total of 98 articles have been collected. The first article was published in 1978 and studied the genotoxic effect of several phytoestrogens using the Ames test (MacGregor and Jurd 1978). As shown in Figure 3, around half of the articles (n ¼ 42; 42.86%) were published during the last 10 years prior to our research, confirming thereby the increase of general interest in phytoestrogens.

Overview of the genotoxic and genoprotective effects of phytoestrogens
Of the 98 studies analyzed in the present review, the genotoxic effect was recorded in 34% of studies whereas the genoprotective effect was reported in 43% of the articles. To note that 9% of studies reported no effects and 14% of works showed both genotoxic and genoprotective results. Although in the current review, only commercially available phytoestrogens were analyzed to avoid the heterogenicity associated with the extraction process from herbal extract, it was still hard to compare results between studies herein; as the same concentration, the same cell, and the same compound were rarely reused. The difference in cell sensitivity, concentration, and the route of administration are factors that may induce conflict.

Phytoestrogens and their mechanism of action
The revised data reported that most studies were with genistein and its metabolites (n ¼ 34; 34.69%), followed by resveratrol (n ¼ 18; 18.37%), daidzein and its metabolites (n ¼ 17; 17.35%), quercetin and its metabolites (n ¼ 17; 17.35%) ( Figure 4). Some articles studied more than one phytoestrogen. The possible mechanism of action (genotoxic and genoprotective effects) of phytoestrogens were summed in Table 2.
No evidence of GEN mutagenicity has been reported in the Ames test (Bartholomew andRyan 1980, McClain et al. 2006), nor in vivo in the mouse and the rat micronucleus (MN) test (McClain et al. 2006). Conversely, GEN induced MN formation in L5178Y mouse lymphoma cells (Boos and Stopper 2000). A loss of a normal chromosome 8 and a chromosomal segment on the short arm of chromosome 9, as well as a gain of extra chromosome 20, were observed in Table 1. Overview of the PICOS eligibility criteria.

Population
Animals and cells are subjected to phytoestrogens. No restriction on the type of animal or cell, age, doses of phytoestrogen, route, and duration of exposure. Intervention Phytoestrogens showed genotoxic or genoprotective effects. Other studies were excluded. Comparison Comparison was based between control and the intervention. Outcomes The primary outcome was mutations, micronucleus formation, DNA breaks, chromosome loss or gain, chromatid exchanges, or DNA repair gene changes. The secondary outcomes were changes in the level of the antioxidant enzymes, proteins, and oxidative stress biomarkers. Studies without primary and secondary outcomes were excluded. Study Studies will be limited to experimental studies. Systematic reviews, meta-analyses, opinions, and editorials will be excluded. MCF-10A breast epithelial cells treated with 1 mM GEN for three months (Kim et al. 2008). GEN and its oxidative metabolite (i.e., 3 0 -OH-GEN) were also considered to be genotoxic agents by inducing DNA breaks in HT-29 human colorectal adenocarcinoma cells measured by the comet assay (Schroeter et al. 2019). The DNA damage induced by bleomycin (antineoplastic drug) was determined by comet and MN assays (R. Lee et al. 2004). Notably, the pretreatment using GEN increased the DNA damage caused by bleomycin in HL-60 human leukemia cells whereas a genoprotective effect was observed in human lymphocytes. Depending on the used cell line, GEN may have genotoxic or genoprotective action toward the DNA. Furthermore, treatment of mice with GEN (5, 10, and 20 mg/ kg) decreased the DNA damage induced by 810 mg/kg antileishmanial meglumine antimoniate measured by the comet and MN assays (de Jesus et al. 2018).
GEN has been identified as an interfacial poison of topoisomerase II (topo II), an enzyme that manages the supercoiling and tangling of DNA by giving rise to transient breaks.
Notably, GEN stabilizes the cleavage complex (topo II-DNA) by intercalating into the DNA at a cleaved scissile bond, thus preventing the occurred strand breaks from religation, leading to clastogenic effects (Ketron and Osheroff 2014). Moreover, GEN is converted to catechols and o-quinones through aromatic hydroxylation. These quinones are covalent poisons that form adducts with topoisomerase I and can redox cycle, resulting in the formation of reactive oxygen  species (ROS) and DNA damage (Pendleton et al. 2014, Bolton andDunlap 2017). The hydroxylated metabolite 3 0 -OH-GEN possessed the same topo II poisoning potential as GEN. Of note, hydroxylated metabolites can generate ROS resulting from oxidation to quinone. Strikingly, 6-OH-GEN showed catalytic inhibitor properties, preventing topo II from producing DNA strand breaks (Schroeter et al. 2019).
On the other hand, the genoprotective effect of GEN is associated with its antioxidant properties by upregulating the gene expression of superoxide dismutase (SOD), glutathione (GSH), and catalase (CAT) which are involved in ROS detoxification (Alfa and Arroo 2019). Further, GEN can enhance the expression of DNA repair genes such as BCRA1, ATM, and p53 (Bhamre et al. 2010, Z. Zhang et al. 2013).

Resveratrol
Resveratrol (RSV, PubChem ID: 445154, molecular formula: C 14 H 12 O 3 ) is found in black grape (0.15 mg/100 g FW) and red wine (from Muscadine grape, 3.02 mg/100 mL) (Rothwell et al. 2013, PubChem 2022. Besides grapes and wine, RSV has been isolated from berries, peanuts, fruits, and vegetables (X. Zhang et al. 2021). RSV has a trans form in food products, but after ultraviolet or visible light exposure, a trans to cis isomerization can occur (Anisimova et al. 2011, Silva et al. 2013. The cis form is not available on the market given its instability (Cottart et al. 2010).
RSV displayed a genotoxic activity in L5178Y mouse lymphoma cells (Schmitt et al. 2002) and induced an increase in chromosome aberrations (breaks and gaps) frequency in wild-type DT40 chicken B cell line after treatment with 50 mM concentration (Liu et al. 2017). RSV also exacerbated sister chromatid exchanges six-fold in the Chinese hamster lung cell line at 10 mg/mL compared to control (Matsuoka et al. 2002). This could be related to the capacity of RSV to bind to the minor groove with a high affinity to AATT and TTAA sequences (Nair et al. 2017). Further, in the presence of Cu (II), a DNA-RSV-Cu (II) complex is formed, allowing the cleavage of DNA (Shaito et al. 2020).
In vitro studies had also suggested that RSV blocks DNA replication or inhibits key enzymes critical for DNA synthesis such as DNA polymerase and topo II. Based on cell-cycle delay and the increased DNA damages, RSV has been believed to be a topo II poison (Demoulin et al. 2015, Liu et al. 2017. However, RSV has been confirmed to act as a topo II inhibitor, able to suppress the cleavage complex formation by preventing the dimerization of the ATPase domain, required for the ATP hydrolysis during topo II activity (J.H. Lee et al. 2017).
It seemed that 50 and 100 mg/kg of RSV had a protective effect against c-radiation-induced genotoxicity (Koohian et al. 2017). Following the MN and chromosomal aberration tests on Allium cepa, the doses of 400 mg/L and 800 mg/L of RSV reduced the MN incidence and all aberration types in cupric chloride (CuCl2) solution (Macar et al. 2020). Although a remarkable protective effect of RSV has been observed while using comet assay on rat C6 astroglial cells exposed to ammonium chloride, a per se 15% DNA damage has been reported (Bobermin et al. 2018). Additionally, results from somatic mutation and recombination test (SMART) and comet assay in Drosophila showed genoprotective potential of RSV against DNA damages evoked by ethyl methanesulfonate and potassium dichromate (Turna et al. 2014). In A549 human lung carcinoma cell lines, a biphasic pattern was observed. At lower concentrations (1 and 5 mM), RSV reduced the sodium arsenite-induced DNA damage, but the damage increased at higher concentrations of the phytoestrogen (20 mM) (Chen et al. 2013).
RSV exhibits a hormetic dose-response effect; at high concentrations, RSV exerts a prooxidant effect, enhancing oxidative stress and thus DNA damage. On the other hand, at lower concentrations, RSV acts as an antioxidant, scavenges the ROS, increases the expression of antioxidant proteins such as deacetylase sirtuin 1 (SIRT1), and thus protects the DNA (Denu 2012). Interestingly, types of cells, time of treatment, and chronobiology can vary RSV effects; RSV exerted prooxidant effects during the light span, whereas in the dark span, RSV acted as an antioxidant (Gadacha et al. 2009, Martins et al. 2014. Generally, the genoprotective effect of RSV is mainly ascribed to the reduction of ROS production, the ability to scavenge free radicals, the stimulation of antioxidant enzymes activities, and the induction of autophagy via the mTOR-dependent pathway (Meng et al. 2020). Enhancement of the expression of genes related to detoxification and DNA repair genes and the suppression of the metabolic activation of promutagens mechanisms were also suggested (Abraham et al. 2015).
On another note, DAI partly induced CREST (þ) (with centromeres/kinetochores due to aneugenic event) and CREST (À) (without centromeres/kinetochores owing to clastogenic event) MN in V79 cells at 100 mM concentration (Kulling and Metzler 1997). However, the MN formation was not induced by DAI up to 100 mM concentration in L5178Y mouse lymphoma cells (Schmitt et al. 2003). DAI has also been reported to act as a catalytic topo II inhibitor with the ability to antagonize GEN for the same binding site on topo II (Snyder andGillies 2003, Bandele andOsheroff 2007).
The genotoxic effect of QUE has been recorded in K-562 human leukemia cells (Das et al. 2017), and in MCF-7 human breast cancer cells at 100 mM concentration (Ragazzon et al. 2009). Interestingly, a recent study revealed both clastogenic and anti-clastogenic effects of QUE in bone marrow cells, depending upon the exposed concentration (Parveen and Shadab 2017).
The DNA damage may be attributed to the high binding efficiency of QUE, owing to the presence of hydroxyl groups in the 3 0 and 4 0 positions on the B-ring (Das et al. 2017). QUE binds to DNA to the G base; the sequence GGGGCCCC is the preferable binding site of QUE (Mitrasinovic et al. 2013, Mitrasinovic 2015. QUE can also inhibit topoisomerase II and poly (ADP-ribose) polymerase (PARP), thus producing DNA double-strand breaks and impeding DNA repair (Biechonski et al. 2017, Raffa et al. 2017. Moreover, QUE can bind to Cu (II) ions to form a complex with DNA. Further, the Cu (II) ions can inhibit CAT activity in the presence of QUE, contributing to the accumulation of ROS and DNA damages (Das et al. 2017, Alcaraz et al. 2021. Strikingly, a genoprotective potential of QUE against urethane and arsenite has been reported (Roy et al. 2008, Nagpal andAbraham 2017). Presumably, this effect is related to the non-enzymatic repair of DNA damage (Tan et al. 2009) or the interference with the ATM pathway of DNA repair (Ye et al. 2004), or just the effect of QUE metabolites (QUE-3-O-glucuronide) and not the QUE per se (Yamazaki et al. 2014).

Kaempferol
Kaempferol (KP, PubChem ID: 5280863, molecular formula: C 15 H 10 O 6 ) is a flavonol, mostly founds in spices such as capers (104.29 mg/100 g FW), cumin (38.60 mg/100 g FW), cloves (23.80 mg/100 g FW), and caraway (16.40 mg/100 g FW) (Rothwell et al. 2013, PubChem 2022. In human lymphocytes and sperm, KP at 100-500 mM concentrations has protected against DNA damage induced by GEN (250 mM), DAI (250 mM), H 2 O 2 (80 mM), diethylstilbestrol (250 mM), and b-estradiol (70 mM) (Cemeli et al. 2004). Rusak et al. (2010) reported the genotoxic effect of KP in human lymphocytes incubated for 2, 4, and 18 h with 1,3 and 10 mM concentrations. The DNA damage recorded at 1 and 3 mM was in the following order 2 h > 18 h > 4 h. However, at 10 mM, the pattern damage of KP increased over time (4 h > 2 h). These findings suggest that the DNA damaging of KP is a time and concentration-dependent effect. At low concentrations, the 4 h exposure period induced less damage compared to the 2 h period due to the efficient DNA repair mechanism. DNA damages were observed after 18 h incubation. This one was assumed to be related to oxidative stress and interference with metabolic processes (Rusak et al. 2010). Overall, a complex KP-DNA is formed in the minor groove thereby yielding DNA damage (Pradhan et al. 2015).

Epigallocatechin-3-gallate
The (À) epigallocatechin-3-gallate (EGCG, PubChem ID: 65064, molecular formula: C 22 H 18 O 11 ) is one of the most abundant catechins present in green tea (infusion, 27.16 mg/ 100 mL) and black tea (infusion, 9.12 mg/100 mL) (Rothwell et al. 2013, PubChem 2022. Johnson and Loo (2000) reported concentration-dependent genotoxicity of EGCG from 100 mM, whereas at 10 mM concentration genoprotective effects toward H 2 O 2 (25 mM) were recorded in human Jurkat T-lymphocytes. Similarly, the concentration of 80 mM had shown a genotoxic effect when applied to human peripheral lymphocytes (Bertram et al. 2003). Surprisingly, the genotoxic concentration observed by Johnson and Loo (2000) was suggested as a genoprotective concentration for human peripheral leucocytes against bleomycin (20 mg/mL) (Glei and Pool-Zobel 2006). Probably, this discrepancy is related to the variation in treatment duration (18 h vs 30 min), the intensity of DNA damage induced by the genotoxic agents (H 2 O 2 , bleomycin), and cell types.
The genotoxic effects of EGCG were ascribed to its intercalation into DNA in the major groove (Galindo-Murillo and Cheatham 2018). There is also an assumption that catechins mobilize the Cu (II) yielding oxidative DNA breakage (Farhan et al. 2015). Probably, the accumulation of EGCG in DNA blocks the protecting effects, since, at low concentrations, EGCG reduces ROS production (Bertram et al. 2003).
LT showed a DNA-damaging effect in human lymphocytes, as well as protective action toward hydrogen peroxidestressed lymphocytes (Rusak et al. 2010). The probable reason is that LT is intercalated into DNA which may stimulate the DNA repair mechanism (Das et al. 2017, Bhuiya et al. 2019).

Apigenin
Apigenin (AP, PubChem ID: 5280443, molecular formula: C 15 H 10 O 5 ) is a flavone abundantly found in olive oil (extra virgin, 1.17 mg/100 g FW) and Italian oregano (fresh, 3.50 mg/ 100 g FW) (Rothwell et al. 2013, PubChem 2022. AP mitigates the genotoxic effect of the fungicide edifenphos in human lymphocytes by reducing ROS production (Ahmad et al. 2019). Studies have demonstrated that AP scavenges free radicals and restores detoxification enzyme activities (Middleton et al. 2000). Interestingly, a recent study concluded that the genoprotective effect of AP against type B ultraviolet (UVB) radiation is associated with its potential to inhibit cyclobutane pyrimidine dimers production, a photoproduct responsible for a variety of genetic mutations, owing to the presence of a double bond between C1-C2 (Garc ıa Forero et al. 2019). Moreover, AP inactivated the metabolism of nitropyrenes, thus reducing the frequency of sister chromatid exchanges in Chinese hamster ovary cells treated with l-nitropyrene (1-NP) or 1,6-dinitropyrene (1,6-DNP) (Kuo et al. 1992).

Equol
Equol (EQO, PubChem ID: 91469, molecular formula: C 15 H 14 O 3 ) possesses a chiral center, thus can occur in two distinct enantiomeric forms: R-equol and S-equol (Rothwell et al. 2013, PubChem 2022. Of note, the enantiomer produced by gut microflora via metabolic reduction from DAI is S-(-)EQO. This EQO enantiomer has a high affinity to bind to estrogen receptors (ER-b) that are present in the epithelium of the prostate, bladder, colon, adipose tissue, and immune system whereas the R-equol has an affinity to bind to ER-a receptors which are found in the mammary gland, uterus, bone, testes, epididymis, the stroma of prostate, liver, and adipose tissue (Muthyala et al. 2004, Setchell et al. 2005, Paterni et al. 2014. A commercially available form named racemic EQO, a mixture of S-equol and R-equol, is also produced from DAI and O-methylated isoflavone known as formononetin (Shinkaruk et al. 2010).
Racemic EQO has shown genotoxic effects in several studies with a pattern resembling the aneugens (Schmitt et al. 2003, Lehmann et al. 2005. However, no genotoxic evidence has been reported for S-EQO. Probably, it is ascribed to the lack of reactive functional groups and the absence of a planar structure to ensure intercalation into DNA (Schwen et al. 2010).

Rutin
Rutin (RU, PubChem ID: 5280805, molecular formula: C 27 H 30 O 16 ) is a flavonoid mostly found in capers (332.29 mg/ 100 g FW) and black olive (raw: 45.36 mg/100 g FW) (Rothwell et al. 2013, PubChem 2022. Studies reported genoprotective effects of RU against H 2 O 2 in human lymphocytes and arsenite in V79 cell lines, which may be ascribed to its antioxidant properties (Noroozi et al. 1998, Roy et al. 2008. Despite the therapeutic potential of this phytoestrogen, RU has a low solubility that may impair its bioavailability (Sharma et al. 2013).

Concentrations and biphasic responses of phytoestrogens
The most used phytoestrogens recorded in this review (genistein, resveratrol, daidzein, quercetin) were selected to assess the relationship between concentration and outcomes (genotoxic, genoprotective). Data were retrieved from the collected articles (Supplementary Table 1). Concentrations were categorized according to the observed outcomes. Based on Table 3, no specific concentration is settled to definite the pattern of actions of the selected phytoestrogens. Notably, the actual effective concentration of these phytoestrogens that need to be supplemented in humans is indefinite. Further, when in vitro effective concentrations can be obtained in vivo is still uncertain.
Some phytoestrogens elicited in vitro bidirectional responses; at low concentrations genoprotective effects and at high concentrations genotoxic effects are observed (Table 4). This phenomenon is known as hormesis, where a cell or an organism reacts to compounds by biphasic responses; stimulation at low concentrations (beneficial effects) and inhibition at high concentrations (harmful effects) (Mattson 2008). In other terms, hormesis is the response of organisms toward an extrinsic or intrinsic stimulus that can involve various signal transduction processes leading to a biphasic response. Thus, the hormetic dose response identifies the limit of biological plasticity needed for the field of pharmaceuticals and natural products to enhance the biological process. In this way, the dose of pharmaceutical/natural products with maximum stimulatory response can be predicted (Calabrese andMattson 2017, Calabrese 2018). The most frequent hypothesis proposed to explain the biphasic pattern was the saturation of defense mechanisms (e.g., detoxification, DNA repair) that may differ across cell types given the difference in the cell cycle control (i.e., p53 functional status) and the DNA repair capacities (Hartwig et al. 2020).

Limitations and future perspectives
In the current review, we evaluated the possible mechanism of action of phytoestrogens ( Figure 5). Herein, studies were selected based on commercially available phytoestrogens to limit the heterogenicity related to the extraction of herbal extracts. Despite this selection, heterogenicity in experimental design was observed, thus making it hard to presume the expected effect from each compound.
Most of the studies discussed in this review used a large range of concentrations, with no standardized concentration neither in vitro nor in vivo. Therein, narrowing down the range to determine the effective concentration is challenging. Moreover, the collected data on the biphasic effect may have implications for therapeutic interventions. However, the in vivo studies were limited to support the in vitro findings. For a more effective extrapolation from in vitro biphasic responses to in vivo, physiologically based pharmacokinetic modeling should be applied, taking into consideration the expected concentrations in tissues (van der Woude et al. 2003).
It is unlikely that humans would be subjected to high concentrations of phytoestrogens. Therefore, low concentrations of phytoestrogens should be tested since the reported  physiological plasma concentrations are in the nM range and occasionally reach the mM concentrations (Rietjens et al. 2017). Another interesting point is that the observed responses obtained may be associated with the experimental design (e.g., cells type tested). Moreover, studies on phytoestrogens mostly focused on short-term exposure. It is worth assessing the long-term exposure that may reveal bioaccumulation or adaptation phenomena. Thus, more feasible strategies should be adopted since a heterogenicity in the study design with different results was reported.
As in vivo tests are costly, time-consuming, and raise ethical concerns, there is momentum to develop more sensitive and alternative methods to evaluate mutagenicity. Further, the existing in vitro methods possess a low specificity (high rate of false-positive) that may lead to unnecessary in vivo follow-up. The improvement of these assays is recommended but without compromising the sensitivity.

Conclusion
This review reports more evidence to support the genoprotective effect of phytoestrogens compared to their genotoxic action. However, it is inconclusive at which concentration the protective role of phytoestrogens can be attained. Hence, further studies are needed to exclude human risk, with respect to concentrations and time exposure used.

Disclosure statement
The authors have no competing interests to declare that are relevant to the content of this article.

Funding
The author(s) reported there is no funding associated with the work featured in this article.