Conceptualizing the Role of the Microbiome as a Mediator and Modifier in Environmental Health Studies: A Scoping Review of Studies of Triclosan and the Microbiome

Abstract

Purpose of Review:

Triclosan is an endocrine disrupting antimicrobial additive that is suspected of contributing to antibiotic resistance and altering the microbiome. In this scoping review, we summarize what is known about the association between triclosan exposure and the microbiome using evidence from in vivo and epidemiologic studies.

Recent Findings:

Our review includes eleven rodent studies, seven fish studies, and five human studies. Evidence from animal studies suggests that triclosan decreases diversity of the microbiome, although only one epidemiologic study agreed. Most studies suggest that triclosan alters the microbial community beta diversity, but disagree on which taxa contributed to compositional differences. Taxa in the Bacteroidetes, Firmicutes, and Proteobacteria may be more influenced by triclosan than those in other phyla.

Summary:

Studies on triclosan and the microbiome were scarce, and were inconclusive as to the effects of triclosan on the microbiome. Additional research is needed clarifying windows of heightened susceptibility of the microbiome to triclosan. We recommend guidelines for future microbiome research in environmental health to increase comparability across studies.

Introduction

Triclosan was developed as a broad-spectrum antibacterial and antifungal compound for use in hospitals, but has been widely added to domestic products, with production increasing 100-fold across the 1990s [1,2]. The U.S. Food and Drug Administration (FDA) banned certain uses of triclosan in 2016 (effective in 2020) due to triclosan’s potential to disrupt hormones and contribute to antibacterial resistance [3]. As of 2019 it was also voluntarily removed from toothpastes in the U.S. However, exposure continues through consumer products not regulated by the FDA including toys, clothing, and kitchenware [4,5] and through its use as a pesticide [6]. Although triclosan is not persistent (half-life < 24 hours) [7], urinary concentrations are stable over time within individuals (ICC = 0.4–0.6) due to repeated exposure [8–10]. Triclosan has been associated with adverse health effects including altered neurobehavior [11–17], decreased cardiometabolic function [18–24], impaired thyroid hormone signaling [25–29], gonadotropin and gonadal hormone imbalances [25,26,30–32], and altered immune and inflammatory responses [33–36]. However, associations are inconsistent across studies and the mechanisms underlying these associations are not fully understood.

The human microbiome may be the key to understanding variability in prior studies as well as uncovering how triclosan impacts human health [37]. Made up of bacteria, archaea, fungi, and viruses that live in and on the human body, the human microbiome performs functions essential to the human host including nutrient and toxicant metabolism and immunomodulation [38]. However, certain pathogenic species may overtake symbiotic and neutral organisms leading to a localized or systemic infectious response. As an antimicrobial, triclosan is suspected to impact the composition and diversity of the gut microbiome, including advantaging species that have evolved antimicrobial resistance [39–42]. Furthermore, human-affiliated bacteria are suspected of reactivating glucuronidated triclosan after host metabolism by deconjugation [43], potentially worsening its effects. Despite a growing literature examining the interaction of triclosan and the microbiome, there has been no review of the literature for comparisons across studies.

In this scoping review, we first broadly characterize the role of the microbiome in environmental health studies. We then provide a review of epidemiologic and in vivo studies examining the association between triclosan and the microbiome. We elected to not include in vitro or in silico studies because these tend to focus on one or a few microbes rather than entire microbial systems, which has limited applicability to human-affiliated microbiomes. Similarly, studies of triclosan in wastewater systems include many aerobic microbes that do not occur in human-affiliated microbiomes. We then close with some recommendations of best practices for future environmental health studies of the microbiome and identify gaps in the literature on triclosan and the human microbiome. Through this review, we hope to synthesize the state of the literature on the effects of triclosan on the microbiome and establish a framework for conceptualizing the role of the microbiome in environmental health studies.

The Microbiome in Environmental Health

In environmental epidemiology, the microbiome may be thought of as a mediator or modifier of associations between environmental contaminants and health outcomes (Figure 1). Through the mediation pathway, contaminants impact the diversity or composition of the human microbiome species or their functional potential and byproducts. These changes then result in differences in health phenotypes across the life course. Indeed, there is growing evidence that environmental contaminants impact the microbiome [44–46] and that the microbiome is related to a range of health effects across the life course, including those related to triclosan exposure [47–49]. However, few epidemiologic studies have explicitly investigated mediation by the microbiome [50,51]. Similarly, few studies have examined effect modification by the microbiome in environmental health studies [52–54]. Modification may occur if a microbial species contributes to or alters host metabolism of an exposure, leading to the production of contaminant metabolites that are more or less toxic than they would be in the absence of the microbe. For example, we found that the infant gut microbiome may contribute to arsenic metabolism [55]. Alternatively, a microbe could confer resilience against certain exposures by priming the host to mount an effective response or induce susceptibility by taxing host response systems [56]. For instance, we found that features of the microbiome including species and gene relative abundances modified the association between arsenic and scores on behavioral measure of autism [52]. It is important to consider that these pathways may co-occur (i.e., bacterial species may act as both a mediator and modifier of a given association); thus epidemiological methods that can address this duality (e.g., four-way decomposition [57]) may be required.

Figure 1.

Conceptual diagram of the potential roles of the human microbiome in environmental health.

In this review, we focus on the association between triclosan and the microbiome because it is an environmental contaminant with the strongest evidence of microbial disruption, and yet, to our knowledge, there has not been a comprehensive review of the literature. In contrast, several reviews of the microbiome and triclosan-related health outcomes have been published in recent years [47–49]. Thus, our goal through this review is to better understand the effect of triclosan on the human microbiome as an intermediate health-related outcome (i.e., the first arrow of the mediation pathway), informed by both epidemiologic and in vivo studies. While many studies of the human microbiome highlight the fecal microbiome, because triclosan was an additive in toothpastes and dermally applied sanitizers, the skin and oral microbiomes may be affected equally or to a greater extent than the those in the gastrointestinal tract. In many cases, studies focus on changes in bacterial composition, but depending on sequencing method changes in fungi, archaea, and viruses may be detected. This context is important for the studies described here and for consideration in the design of future microbiome studies.

Evidence from Animal Studies

Studies in Fish Models

Across the collection of fish studies investigating the impact of triclosan exposure on the microbiome, several consistent trends emerge (Table 1). The studies employ a variety of model systems, including zebrafish and minnows, and assess triclosan exposure through different routes, such as food, water, and culture dishes. In general, triclosan exposure appeared to substantially decrease the alpha (within-subject) diversity of the microbiome [58–61], although in one case triclosan exposure was found to be associated with an increased alpha diversity [62]. Two studies found no association between alpha diversity and triclosan exposure [63,64]. Notably, the addition of certain interventions, like probiotic diets, appeared to mitigate the reduction in diversity caused by triclosan exposure [61]. Beta diversity (between subject diversity) assessments reveal that triclosan exposure is associated with significant changes in microbial community composition. Exposure groups cluster separately from control groups and display distinct beta diversity patterns in all fish studies included in this review.

Table 1.

Observed modulations in the relative abundance of gut bacterial after triclosan exposure in fish studies (n=7)

	Significant Increase	Significant Decrease	No Change*
Alpha Diversity	1	4	2
Firmicutes	0	1	6
Proteobacteria	1	2	4
Bacteroidetes	0	1	6

^*May indicate no statistically significant change or no mention of a significant change

Differential abundance and relative abundance analyses highlight specific taxonomic shifts in response to triclosan exposure. For example, Firmicutes, Proteobacteria, and Bacteroidetes are frequently mentioned phyla affected by triclosan exposure. However, these impacts are not consistent across all studies, suggesting that the effect of triclosan on specific taxa within these phyla can be complex. Moreover, the direction of the impact for the taxa mentioned, whether there was an increase or decrease in bacterial abundance, is also inconsistent across the studies. For example, in a 2021 study looking at the impacts of triclosan on zebrafish larvae, the relative abundance of Proteobacteria decreased in 0.03, 0.3, 30, and 300 ng/ml triclosan exposure groups, but increased in the 100 ng/ml triclosan group [59]. In contrast, in a 2019 study of adult Zebrafish, triclosan exposure led to a significantly higher abundance of Proteobacteria [61].

Notably, one 2015 study showed that when the fish were followed after triclosan exposure for two weeks, their microbiomes were not distinguishable from the non-triclosan exposed groups, suggesting that the effects of triclosan may not be sustained over time [62]. Additionally, a common finding across these studies that tested microbial-related interventions (including probiotics and short-chain fatty acids) is their potential to mitigate the impacts of triclosan exposure on diversity and composition [60,61]. This suggests that manipulating the microbiome through dietary or other means might offer a strategy for addressing triclosan-induced disruptions.

Overall, while there are variations in the specific findings among these studies, the consistent themes point to the potential for triclosan exposure to impact fish microbiomes, altering diversity and composition in a dose-dependent and time-dependent manner. Additionally, the use of interventions to mitigate these effects suggests that strategies to restore microbiome balance could be explored as potential protective measures in response to triclosan exposure.

Studies in Rodent Models

Previous studies have used animal models to assess the compositional and functional impact of triclosan exposure on the gut microbiome. Eleven studies investigated whether environmentally relevant levels of triclosan induced perturbations in the gut microbiota of rodents. All eleven rodent studies examined differences in alpha diversity, beta diversity, and the relative abundance of certain bacterial taxa after triclosan exposure (Table 2). Seven studies reported a statistically significant decrease in alpha diversity [20–22,43,65–67], while three papers cited no marked changes in species richness after triclosan exposure [68–70]. The observed reduction in alpha diversity may reflect alterations in more vulnerable microbial communities, as species richness is critical for bolstering resilience within the gut microbiome [66]. Notably, one study found that triclosan treatment increased species richness and restored the decreased bacterial diversity in mice on a high-fat diet [71]. Taken together, these findings suggest that triclosan exposure is generally associated with a reduction in the alpha diversity of rodent gut microbiomes.

Table 2.

Observed modulations in the relative abundance of gut bacterial after triclosan exposure in rodent studies (n=11)

	Significant Increase	Significant Decrease	No Change*
Alpha Diversity	1	7	3
Lachnospiraceae	2	2	0
Clostridiaceae	2	3	0
Firmicutes	1	4	1
Bacteroidetes	5	2	1

^*May indicate no statistically significant change or no mention of a significant change

Triclosan treatment was also associated with modulations in beta diversity, as evidenced by distinct microbial structures between control and treatment groups [21,34,43,65–67,69,70]. These marked changes in beta diversity may indicate shifts in the relative abundance of certain bacteria [67] or reductions in overall microbiota diversity [21]. Notably, one study found that low dose triclosan treatment resulted in statistically significant changes to gut microbiota diversity in adolescent rats, while no such difference was recorded in adult rats of the same breed [70]. This finding denotes the possibility of a developmental window during adolescence when the rodent gut microbiome is particularly susceptible to triclosan exposure. Of the eleven studies, two found no significant modulation in microbial beta diversity [68,71], while one paper made no explicit mention of beta diversity results [20]. Across the selected rodent studies, various statistical indices were employed to measure microbial diversity, including the Principal Coordinate Analysis [20,34,43,65,66,69,71], Bray-Curtis dissimilarities [65,68,70], Nonmetric Multidimensional Scaling [65,71], and Weighted UniFrac Distances [43,67,69]. This may partly explain the observed discrepancies in beta diversity findings between studies. More specifically, these variations in beta diversity parameters may highlight the potential for differences in data interpretation and outcomes, which warrant further consideration when comparing the results yielded across experiments.

Across the summarized studies, differential abundance of intestinal bacterial taxa was analyzed using a variety of methods and with varying levels of reporting specificity. For the purposes of this review, we focused on the bacterial clades that were consistently mentioned across multiple papers: the Lachnospiraceae and Clostridiaceae families, and Firmicutes and Bacteroidetes phyla. The changes in relative abundance for both Lachnospiraceae and Clostridiaceae are inconsistent across the rodent studies. Hence, further research would be useful for clarifying how triclosan treatment affects the relative abundances of these gram-positive anaerobes. Most studies reported a statistically significant decrease in relative abundance of Firmicutes following triclosan exposure [67,70–72], while one paper cited a marked increase [21]. In contrast, Bacteroidetes generally increased with triclosan exposure across studies [20,65,67,70,71], with only two papers reporting a notable decrease [21,72]. Collectively, these results suggest that triclosan exposure is associated with significant decreases in Firmicutes and increases in Bacteroidetes.

Comparability between studies is limited by methodological differences. For example, two studies analyzed cecal microbiome composition in contrast to the others, which used fecal contents. For the two studies that utilized cecal contents as the DNA source, triclosan exposure was not associated with significant changes in alpha diversity nor differences in the relative abundance of Bacteroidetes and Firmicutes [68,69]. Previous research has highlighted several differences between the relative abundance of certain bacteria found in fecal versus cecal samples [73]. Hence, it is plausible that the observed discrepancies between studies can be partly attributed to differences in the DNA source (i.e. which portion of the gastrointestinal tract was utilized for extraction) in addition to differences in dosage, timing, and mechanism of exposure and baseline microbiome composition.

Triclosan-induced shifts in the abundance of Firmicutes and Bacteroidetes, amongst other bacteria, may contribute to the development of intestinal dysbiosis. Recent experimental evidence has associated intestinal dysbiosis with the emergence of adverse health outcomes in rodents. For instance, a 2022 study concluded that triclosan treatment exacerbated experimentally induced colitis in Balb/c mice [65]. This finding parallels the results of an earlier study, which found that a three-week exposure of triclosan increased the severity of colonic inflammation and colitis-associated colon tumorigenesis in Tlr4 −/− mice [34]. Triclosan treatment has also been associated with notable liver injuries characterized by inflammation [68,72] and bile acids metabolism disorders in C57BL/6 mice [68]. Despite the inherent limitations of utilizing rodent models, these findings can begin to elucidate the mechanisms through which triclosan exposure alters the species diversity and relative abundance of bacteria in the human gut microbiome.

Evidence from Human Studies

Five human studies examined associations between triclosan and the microbiome in humans, two of which studied oral samples [74,75] and four of which studied stool samples [75–78], including one which studied both oral and stool microbiomes [75]. Of the four human studies that discussed alpha diversity, three found no associations between alpha diversity of the microbiome and triclosan [74,75,77]. One study did find that the alpha diversity, when measured by Shannon’s Diversity Index, was statistically significantly decreased in infants who were exposed to triclosan through human milk compared to those who were not; however, there was no difference when using the Chao1 Index [76]. This finding may highlight a potential vulnerability of the microbiome during early life to triclosan. Similarly, beta diversity analyses did not show associations with triclosan. Only three studies were found to investigate the relationship between triclosan and the beta diversity of the microbiome, and none of them found statistically significant associations.

When analyzing microbial taxonomic abundance and relative abundance, the human studies indicated that triclosan exposure had limited effects on specific bacterial taxa. While some associations between certain taxa and triclosan exposure were identified, in general, the overall taxonomic composition did not show consistent or pronounced shifts because of exposure. The majority of human studies did not find alterations to abundance due to triclosan [74,75,78]. In contrast, the study investigating the presence of triclosan in breast milk and its impact on the infant microbiome found statistically significant associations [68]. Infants that were exposed to detectable levels of triclosan in their human milk had higher relative abundance of the family Rhodospirillaceae [68]. Additionally, a randomized control trial in 2017 found that introducing mothers and infants to triclosan-containing personal care products within the first year postpartum increased the relative abundance of Proteobacteria, which is a broadly antibiotic-resistant phylum [67]. The same study also found that the microbiomes of infants with lower triclosan levels were enriched with Bacteroides fragilis, which was not found in the microbiomes of their mothers [77]. Both studies consistently suggest a potentially increased vulnerability to triclosan with the infant microbiome compared to adults.

Despite the statistically significant differences observed in some studies, particularly in infants, the consensus, even in studies that did find associations, was that triclosan exposure has modest effects on the microbiome, with these effects often not persisting over time. The studies collectively suggest that triclosan exposure is not strongly associated with major shifts in microbial diversity or community structure within the studied populations. However, limitations are evident in many of the studies, including low sample sizes and only a subset of samples within each of the studies containing triclosan levels above the limit of detection. A cross sectional study from 2022 that examined the impacts of triclosan on the gingival microbiome (n=477) found that triclosan levels were above the limit of detection only 26% of the time (n=123) [66]. Another cross-sectional study from 2018 that investigated the impact of triclosan in human milk on the infant microbiome had a sample size of 45, and only 27% of samples detected triclosan above the limit of detection (n = 12) [68]. The variability in exposure levels, often in already small sample sizes, could potentially impact the ability to detect statistically significant associations. Additionally, each study included a different set of confounders, often over-adjusting given the study design and statistical power, which may have led to null results and limited comparability of estimates across studies. The limitations and inconsistency of results suggest further research is required to fully determine the impact of triclosan on the human microbiome.

Challenges and Best Practices for Studying the Microbiome in an Epidemiologic Setting

In reviewing the literature of how triclosan relates to the microbiome, we identified several gaps that are likely generalizable to the microbiome in environmental epidemiology more broadly. First, there has been little effort to identify windows when the human microbiome may be particularly susceptible to triclosan exposure. For example, there is evidence that the microbiomes of infants and adolescents may be more sensitive to environmental exposures [70,79]. However, it is unknown whether exposure in these vulnerable windows has lasting implications for the microbiome (e.g., does infant triclosan exposure impact the childhood microbiome). Future studies with multiple windows of exposure may inform our understanding of the persistence of microbiome shifts. Second, the variety of methods for quantifying the microbiome means comparability between studies is limited. Bias and/or differences may arise in sample collection and storage, DNA extraction, sequencing method, and read processing pipelines. For this reason, it is essential that each of these steps is rigorously documented and reported so that researchers know whether results are directly comparable or whether there are factors that may explain variability in results. Finally, more advanced statistical methods are needed to handle advanced epidemiologic approaches (e.g., mediation) while accounting for the zero-inflation and autocorrelation of microbiome data. For epidemiologic studies, this includes methodologies that are better able to handle complex confounding.

It should be acknowledged that humans are not exposed to triclosan in isolation. One prior rodent model examining triclosan alone and in a mixture with other endocrine disrupting compounds that may have a common source found that the mixture had unique effects compared to its component parts, suggesting potential biological interactions between chemicals [70]. Novel mixture methods such as weighted quantile sum regression and Bayesian kernel machine regression have been applied to other chemical classes in association with the microbiome [80,81] and considering mixtures of bacteria associated with exposures or outcomes [82,83]. However, to our knowledge, no studies have examined joint effects of triclosan and other chemicals on the human microbiome. These approaches are limited in their ability to handle the microbiome data structure and more attention is needed to developing appropriate methods for investigating mixtures and the microbiome.

When linking exposure-related differences in the microbiome to health outcomes, measures of the functional capacity of the microbes can be useful to assess biological plausibility (e.g., gene relative abundance, metatranscriptomics, metabolomics). All human studies and all but one [66] of the animal studies included in our review utilized 16S rRNA sequencing as the primary sequencing method, which limits inference of bacterial function. Gao and colleagues, who employed metagenomic sequencing found increases in gene pathways related to stress response, antibiotic resistance, and metal resistance with triclosan exposure. One human study conducted metagenomic sequencing in a small subset of samples, but only aligned sequences to a catalog of antibiotic resistance genes [77]. The authors of this study acknowledge that their null findings are likely due to the small sample size. Two murine and one zebrafish studies used software to infer functional gene relative abundance (e.g., PICRUSt) [64,71,72]. However, these tools have been shown to perform poorly, particularly for non-human microbiomes (e.g., rodents or fish) [84]. One murine study performed and reported on metabolomic differences related to triclosan exposure, but did not connect these differences to changes in the microbiome [68]. As the field advances, more attention to conducting and harmonizing these types of data will be necessary to fully understand the biological pathways from exposures to health outcomes.

While the methodologies used in each environmental epidemiologic study of the microbiome will vary depending on the research question, we recommend certain best practices be followed to facilitate advances in the field. While these are essential in scientific research more generally, they bear repeating in the context of microbiome studies. When designing the study, researchers should ensure their methods align to their research question. For example, if the primary route of exposure to the toxicant of interest is dermal, it is possible the skin microbiome bears the brunt of exposure, protecting internal microbiomes. Similarly, if there is evidence the toxicant of interest has antifungal properties, 18S rRNA or ITS sequencing may be a more appropriate sequencing strategy than 16S rRNA (only in prokaryotic organisms) or metagenomic sequencing (where reads are typically dominated by bacteria). As mentioned previously, it is essential that all choices in sample collection, DNA extraction, sequencing method, and read processing pipelines are documented and reported in all manuscripts. This includes reporting versions of software and algorithms used, parameters selected for analysis (e.g., which estimate of alpha or beta diversity), units for measures of association, and the sample size for each analysis. Additionally, all results should be reported, even if null. Analysis of differential species or gene relative abundance can lead to hundreds or even thousands of statistical tests. While not all of these may be reported in the main text, effect estimates and estimates of variance should be included in the supplemental material to enable future meta-analyses. Addressing the identified gaps and implementing the recommended best practices in future microbiome studies within an epidemiologic setting will be crucial for advancing our understanding of triclosan’s impact on the microbiome and similar environmental exposures.

Conclusions

While studies suggest that triclosan exposure may lead to minor associations with specific taxa, major changes to the microbiome due to triclosan remain unproven. The lack of studies on the topic and inconsistencies across the approaches and findings presented here underscore the complexity of the relationship between triclosan exposure and the microbiome, warranting further investigation to better understand these associations. Further, there is a need for more unified methodological practices to glean a more robust understanding of how triclosan exposure alters the gut microbiome.