A Review on the 40^th Anniversary of the First Regulation of Drinking Water Disinfection By-products

Abstract

Water disinfection, primarily by chlorination, is one of the greatest achievements of public health. However, more than half a century after its introduction, studies in the 1970s reported that (a) chlorine interacted with organic matter in the water to form disinfection by-products (DBPs); (b) two DBPs, chloroform and bromoform, both trihalomethanes (THMs), were rodent carcinogens; (c) three brominated THMs were mutagenic; in six studies chlorinated drinking waters in the U.S. and Canada were mutagenic; and (d) in one epidemiological study there was an association between bladder cancer mortality and THM exposure. This led the U.S. Environmental Protection Agency to issue its first DBP regulation in 1979. Forty years later, >600 DBPs have been characterized, 20/22 have been shown to be rodent carcinogens, >100 have been shown to be genotoxic, and 1000s of water samples have been found to be mutagenic. Data support a hypothesis that long-term dermal/inhalation exposure to certain levels of the three brominated THMs, as well as oral exposure to the haloacetic acids, combined with a specific genotype may increase the risk for bladder cancer for a small but significant population group. Improved water-treatment methods and stricter regulations have likely reduced such risks over the years, and further reductions in potential risk are anticipated with the application of advanced water-treatment methods and wider application of drinking water regulations. This 40-year research effort is a remarkable example of sustained cooperation between academic and government scientists, along with public/private water companies, to find answers to a pressing public health question.

INTRODUCTION

The year 2019 marked the 40^th anniversary of the issuance by the U.S. Environmental Protection Agency (U.S. EPA) of its first regulation on the allowable concentrations of disinfection by-products (DBPs) in drinking water (U.S. EPA 1979). This regulation was developed based on emerging evidence that (a) DBPs were present in chlorinated drinking water, (b) two DBPs (chloroform and bromoform) were rodent carcinogens, (c) chlorinated drinking water was mutagenic, and (d) there was an association between bladder cancer mortality and exposure to a class of DBPs (trihalomethanes) in drinking water. Thus, it seems fitting to revisit these initial findings that prompted the first DBP regulations in the world and to examine how the ensuing science has furthered our understanding of this issue.

Chlorination to disinfect water for drinking and other purposes began sporadically in Europe, notably Germany, Belgium, and England, in the late 1800s; however, widespread use of water chlorination began in the early 1900s (McGuire 2013). Chlorination on a municipal basis began in the U.S. in 1908 in Jersey City, New Jersey, and the impact since its introduction in the U.S. is evidenced by a reduced incidence of cholera by 90%, typhoid by 80%, and amoebic dysentery by 50% (Ohanian et al. 1990). By the year 2000, no cases of typhoid attributable to public drinking water were reported in the U.S. (Levin et al. 2002).

However, as discussed below, evidence began emerging in the 1970s showing that there was likely a trade-off between the elimination of microbial contaminants and the concomitant protection of public health afforded by chlorination, and the formation of disinfection by-products resulting from the interaction of chlorine with organics in the water. Disinfectants such as chlorine are strong oxidants, oxidizing naturally occurring organic matter, anthropogenic contaminants, and bromide and iodide present naturally in most source waters (rivers, lakes, and groundwaters) (Richardson et al. 2007). Until the 1970s this reaction was unrecognized, and the potential health effects of the resulting DBPs were unknown.

As noted below, several early reviews (Loper 1980; Wilkins et al. 1979) have described the initial and subsequent studies on DBPs and the association between exposure to chlorinated water and the risk for cancer. This brief review does not explore these issues again in depth. Instead, this review highlights some of the major findings leading to the 1979 DBP regulation and provides a brief assessment of the situation 40 years later. In addition, some of the key discoveries are noted that have helped to unravel the complex nature of the ubiquitous environmental exposures most of us encounter daily with disinfected water and the potential of such exposures to cause cancer.

EVIDENCE FOR THE LINK BETWEEN EXPOSURE TO DRINKING WATER AND CANCER AT THE TIME OF THE 1979 THM RULE

As reviewed in detail by Loper (1980) and Wilkins et al. (1979), by 1979 there were altogether ~100 papers on the (a) presence of carcinogens in raw water or DBPs in drinking water; (b) mutagenicity of drinking water; (c) carcinogenicity of two DBPs (chloroform and bromoform); and (d) epidemiology showing cancer risk with the use of contaminated raw water to make drinking water, chlorinated versus non-chlorinated (well) water, or exposure to trihalomethanes (THMs) in drinking water. Some of the main findings prior to the 1979 regulations are discussed briefly here and cover the five topics for which there were data relevant to the issue of drinking water and cancer at the time: the detection of carcinogens in source and drinking waters, the detection of DBPs in drinking water, the carcinogenicity of DBPs in rodents, the mutagenicity of DBPs and organic extracts of drinking water, and the carcinogenicity of water in humans.

Carcinogens and Contaminants in Natural Waters and Drinking Water

By the 1940s, aqueous and atmospheric discharges of factories and runoff from oiled roads were considered possible sources of environmental carcinogens that could be contaminating natural waters, such as oceans, rivers, lakes, and streams (Hueper 1942). Newly developed analytical methods revealed that natural (i.e., raw) water and drinking water contained detectable amounts of organic compounds, such as polycyclic aromatic hydrocarbons (PAHs), phenols, insecticides, benzene, etc. (Braus et al. 1951; Middleton and Rosen 1956). The first assessment of carcinogenicity in rodents associated with natural waters was the demonstration that fractionated organic extracts of barnacles removed from wooden pilings off the coast of California were carcinogenic after subcutaneous injection into mice (Shimkin et al. 1951), and the likely carcinogens were PAHs (Shimkin et al. 1951; Zechmeister and Koe 1952). The authors assumed that PAHs were not natural metabolites and would not be expected to be present in organisms; thus, the authors suggested that the PAHs likely came from various sources contaminating the coastal waters.

Additional studies found that organics extracted by activated carbon from a body of water into which oil-refinery effluents were discharged were carcinogenic in rodents when applied to the skin of mice (Hueper and Ruchhoft 1954). A similar study showed that organic extracts of water from the Kanawha River near Nitro, West Virginia, U.S., downstream from large chemical factories that released their effluents into the river, also caused skin cancers in mice when applied to the skin and caused leukemia when injected subcutaneously (Hueper and Payne 1963). Throughout the 1960s and 1970s, studies of natural waters from North America and Europe began documenting the presence of contaminants, including anthropogenic (mostly PAHs) and synthetic organics, some of which were known or suspected rodent carcinogens (Hueper 1961a; Hueper and Payne 1963; Ilnitsky and Varshayskaya 1964; Kraybill 1976; Ott et al. 1978; Wilkins et al. 1979). More broadly, concerns were raised about the increasing number of synthetic chemicals being introduced into the environment with carcinogenic potential (Hueper 1961b).

Chemical analyses of drinking water in the 1960s and 1970s revealed the presence of pesticides, PAHs, and a wide variety of organic compounds (Benoit et al. 1979; Kleopfer 1972; LeBel et al. 1979; McCabe et al. 1970; Schafer et al. 1969; Scheiman et al. 1974). These included various known or suspected human carcinogens (vinyl chloride, benzene, benzo[a]pyrene), as well as known or suspected animal carcinogens (various pesticides, carbon tetrachloride, chloroform) (Safe Drinking Water Committee 1977).

DBPs in Drinking Water

The presence of DBPs in chlorinated water was first reported by Rook (1974) in the Netherlands who identified the formation of four THMs (THM4), i.e., chloroform, bromodichloromethane (BDCM), dibromochloromethane (DBCM), and bromoform, in chlorinated water prepared from the Rhine and Meuse Rivers. He also identified other chlorinated compounds, such as methylene chloride, methylene bromide, 1,2-dichloroethane, and carbon tetrachloride in these chlorinated waters. Rook suggested that these compounds formed via the chlorination of humic substances that occur naturally in source water (i.e., natural waters, which are also called raw waters, used to making drinking water) and that their concentrations were associated with the color of the source water, i.e., the darker the source water, the higher the concentrations of DBPs in the chlorinated water from the source water. Bellar and Lichtenberg (1974) extended these studies by determining the concentrations of chloroform, BDCM, and DBCM in various municipal waters in the U.S. and from a variety of points within a water-treatment plant. They concluded that these DBPs likely formed from the reaction of chlorine with organics in the source water.

The most comprehensive survey of THM concentrations that was made prior to the 1979 THM regulation was conducted by the U.S. EPA as part of the National Organics Reconnaissance Survey (Symons et al. 1975). This survey determined the concentrations of the four THMs, 1,2-dichloroethane, and carbon tetrachloride in drinking water from 80 cities across the U.S. All cities chlorinated their water except for one that used ozonation. Halogenated organics were present in all chlorinated water samples evaluated, and the concentrations determined by this survey were among the data used to set the THM regulatory levels in 1979. The THMs were essentially absent in raw source waters at the levels of detection, which was 1 μg/liter (Bellar and Lichtenberg 1974).

Carcinogenicity of DBPs in Rodents: Chloroform and Bromoform

Before chloroform was known to be a DBP, Eschenbrenner and Miller (1945) at the U.S. National Cancer Institute (NCI) showed that repeated oral administration of chloroform to mice induced hepatomas; their study indicated that induction of necrotic tissue was required for the formation of the tumors. More than 30 years later, Page and Saffiotti (1976), also at the NCI, performed a similar study using repeated oral administration of chloroform and showed that chloroform induced kidney tumors in male rats and liver tumors in male and female mice (NCI 1976). Significant frequencies of these tumors were found even at the lowest doses tested (90–100 mg/kg body weight), with chloroform given by gavage five times per week for 78 weeks.

In a screening study of 16 organic contaminants of U.S. drinking waters, Theiss et al. (1977) found that only bromoform (but not chloroform) induced lung adenomas in strain A mice after multiple intraperitoneal injections. Thus, two DBPs had been shown to be rodent carcinogens by 1979. Considering the structural similarities of these DBPs, the results raised the possibility that the other DBPs might also be carcinogenic. Consequently, studies were initiated to examine the potential health effects of various DBPs (Jolley 1978). The U.S. EPA noted in 1977 that the chemical reactivity of brominated compounds would likely be greater than that of chlorinated compounds like chloroform, and that given the structural similarities of such compounds, brominated THMs would be included in the forthcoming regulation (EPA Statement 1977).

Mutagenicity of Drinking Water and DBPs

Organic extracts of drinking water from five U.S. cities were selected for assessment for mutagenicity in Salmonella, cell transformation in mammalian cells, and carcinogenicity and teratogenicity in rodents (Tardiff et al. 1975). These and related studies were contingent on the development of reliable methods for preparing organic extracts of sufficient concentration for in vitro testing. These included solid-phase extraction of organics from drinking water using XAD resins (Burnham et al. 1972; Chriswell et al. 1977), along with concentrates made by liquid-liquid extraction of reverse-osmosis concentrates (Kopfler et al. 1977).

U.S. and Canadian drinking waters were the first shown to be mutagenic in Salmonella (Chriswell et al. 1978; Glatz et al. 1978; Hooper et al. 1978; Loper et al. 1978; Nestmann et al. 1979; Simmon et al. 1977). The extracts were generally most potent in the base-substitution strains TA1535 and TA100 without metabolic activation (S9); addition of S9 generally reduced the mutagenic potency of the extracts. An extract of drinking water from New Orleans, Louisiana, also induced mammalian cell transformation in BALB/3T3 cells (Loper et al. 1978), indicating that the extract had potential carcinogenic activity. One study showed that ozonation of some source water did not result in mutagenicity (Cotruvo et al. 1978). Although another study showed that ozonation did produce mutagenic water, the water had been drawn from a well near a pond containing treated municipal wastewater (Gruener 1978).

Only one study was published on the mutagenicity of DBPs by 1979 (Simmon et al. 1977), which showed that three brominated THMs (BDCM, BDCM, and bromoform) were mutagenic in Salmonella TA100 when the cells were exposed to the vapors in a desiccator. In contrast, chloroform, which was carcinogenic in rodents only by repeated dosing causing necrosis, was not mutagenic in either the standard plate-incorporation assay or when the cells were exposed to the vapors in a desiccator. This finding confirmed the need by the U.S. EPA to include the brominated THMs along with chloroform in its 1979 THM regulation (EPA Statement 1977).

Carcinogenicity of Drinking Water in Humans

By the mid-20^th century, epidemiological studies in the Netherlands and England indicated that cancer mortality was higher in geographic areas served by surface rather than ground water (Tromp et al. 1955), implying that contaminants in the surface water that were most likely not in the ground water were the cause of the increased cancer mortality. A decade later, Cook and Watson (1966) found that counties in the U.S. State of Missouri, with the highest incidence of multiple primary tumors clustered along the Missouri River, suggesting that carcinogens in the river might be causing the excess cancers.

One of the studies that stimulated additional analytical studies, epidemiology studies, and, ultimately, regulatory action, was the so-called “New Orleans Drinking Water Controversy” (DeRouen and Diem 1975; Page et al. 1976; U.S. EPA 1974, 1975; Wilkins et al. 1979). As reviewed by the authors noted above, the possible association of cancer with carcinogens in New Orleans, Louisiana, drinking water created much controversy, discussion, and public awareness of chemical contaminants in the water supply and the potential health effects exposure to such water might cause.

As reviewed by Wilkins et al. (1979), many of the 19 epidemiological investigations in the U.S. during the 1970s began finding associations between cancer mortality or elevated cancer risk and (a) the presence of chemical contaminants/carcinogens in source water, (b) use of surface water as opposed to ground water, (c) use of chlorinated as opposed to unchlorinated water, (d) increased amounts of chlorination, and (e) the concentration of DBPs (chloroform or THMs) in drinking water. Although most of the associations were generally weak and not conclusive, they were suggestive and warranted further research (EPA Statement 1977).

Regulation of DBPs

The emerging evidence throughout the 1960s of various types of chemical contamination of the air, soil, water, and food led to a call for regulatory action and a systematic assessment of the situation as exemplified by Epstein (1970). Notably, this coincided with the establishment of the U.S. EPA and the Environmental Mutagen Society (now the Environmental Mutagenesis and Genomics Society) (DeMarini 2020). Soon after DBPs were discovered in drinking water, the Safe Drinking Water Act was enacted in 1975, which required the U.S. EPA to “conduct a comprehensive study of public water supplies and drinking water sources to determine the nature, extent, sources of and means of control of contamination by chemicals or other substances suspected of being carcinogenic” (Safe Drinking Water Act 1975).

As reviewed by Larson (1989) and Levin et al. (2002), most states had not adopted or enforced the limited federal drinking water standards prior to 1974. These included standards going back to 1914 on acceptable concentrations of coliform bacteria, and to 1942 on concentrations of arsenic. However, during the 12 years following the enactment of the Safe Drinking Water Act in 1975, the U.S. EPA adopted most of the ~30 previous U.S. Public Health guidelines as interim standards and adopted the regulations for THMs in 1979, making these standards enforceable in U.S. public water systems (Laron 1989; Levin et al. 2002).

The 1979 THM regulation established that the combined concentration of the four THMs (chloroform, BDCM, DBCM, and bromoform) should not exceed an annual average of 100 μg/L in drinking water (U.S. EPA 1979). However, these regulations applied only to municipal water supplies serving >10,000 people that disinfected their water (mainly by chlorination) and not to private wells or other non-public sources of drinking water. This instituted monitoring of THM concentrations by municipal water-treatment facilities to assure that the water they produced met this DBP regulation.

EVIDENCE FOR THE LINK BETWEEN EXPOSURE TO DRINKING WATER AND CANCER TODAY

Below are summarized the advances in the various topics noted earlier, along with additional topics that emerged during the past 40 years. These include the genotoxicity of source water, swimming pool water, and hot tub water; the genotoxicity of swimming; the molecular epidemiology of THM exposure and bladder cancer risk; and a hypothesis for bladder cancer risk associated with THM exposure. The general summary of the field today (below) shows how much has been learned in the last four decades and identifies some emerging issues that require research.

Detection and Analysis of Disinfected Waters for DBPs

Since the discovery of the first DBPs by Rook (1974), >600 mostly low-molecular weight, semivolatile or volatile DBPs have been identified, accounting for ~30% of the total organic halogen (TOX) in some chlorinated waters (Richardson 2011), and the THM4 and haloacetic acids (HAAs) each account for ∼10% of TOX (Krasner et al 2006). Nonetheless, the number of DBPs characterized has increased more than 100-fold since 1974 and will likely continue to increase as analytical chemical methods improve, and more varied types of disinfected waters are assessed.

Although initial studies 40 years ago identified only a few chemical classes of DBPs (THMs and haloacetic acids), a vast array of chemical classes of DBPs have been identified, including haloacetic acids, halonitromethanes, halofuranones, haloamides, haloacetonitriles, nitrosamines, aldehydes, oxyhalides, halopyrroles, haloketones, and halobenzoquinones (Li and Mitch 2018; Richardson et al. 2007; Richardson and Ternes 2018). Some species or classes appear to be linked to components in the source water and/or to treatment methods, and many show complex relationships among each other (Ersan et al. 2019; Li and Mitch 2018).

Carcinogenicity of DBPs in Rodents

Since the first studies showing chloroform and bromoform were rodent carcinogens, a total of 22 DBPs have been evaluated in rodent cancer bioassays (Table I), and all but two, chloroacetic acid and chlorite, were found to be carcinogenic (NTP 2010, 2015; Richardson et al. 2007). In addition, various nitrosamines, which also can be formed during water disinfection, are rodent carcinogens (Richardson et al. 2007). Eight of the 22 DBPs and one nitrosamine exhibit the toxicological features of the majority of human carcinogens in that they are carcinogenic in two or more species and/or are carcinogenic in one species of rodents, and they are genotoxic across a variety systems/endpoints (Baan et al. 2019). These are BDCM, bromate, dichloro- and dibromoacetic acid, bromate, formaldehyde, acetaldehyde, mutagen X (MX; 3-chloro-4-(dichloromethyl)-5-hydroxy-5H-furan-2-one), and nitrosodimethylamine (NDMA). In general, the brominated DBPs are more carcinogenic than the chlorinated ones (Richardson et al. 2007).

Table I.

DBPs Evaluated for Carcinogenicity in Rodents

DBP	Resulta
Trihalomethanes
Bromodichloromethane	+
Bromoform	+
Chlorodibromomethane	+
Chloroform	+
Haloacetic Acids
Chloroacetic acid	−
Bromoacetic acid	+
Dibromoacetic acid	+
Bromodichloroacetic acid	+
Bromochloroacetic acid	+
Dichloroacetic acid	+
Dibromochloroacetic acid	+
Trichloroacetic acid	+
Aldehydes
Acetaldehyde	+
Chloroacetaldehyde	+
Formaldehyde	+
Others
Bromate	+
Chlorite	−
Chloral hydrate	+
Chlorate	+
Chloropicrin	+
Dibromoacetonitrile	+
MX	+

^aDetailed data for all but two are in Richardson et al. (2007); detailed data for bromodichloroacetic acid in NTP (2015) and for dibromoacetonitrile in NTP (2010).

Although chloroform was the first DBP shown to be a rodent carcinogen as noted earlier, it is not mutagenic and requires repeated toxic doses causing necrosis in order to be carcinogenic in rodents. Consequently, chloroform has been evaluated as only a possible (2B) human carcinogen (IARC 1999). Nagano et al. (2006) found that combined inhalation and oral exposures to chloroform in rats produced renal tumors, whereas the separate exposure did not produce such tumors.

Of the DBPs tested for carcinogenicity in rodents, only bromoform, BDCM (Richardson et al. 2007), and bromochloroacetic acid (NTP 2015), all of which are brominated DBPs, induced tumors of the large intestine. This is of mechanistic interest because colorectal cancer, along with bladder cancer, is associated with exposure to disinfected water in humans (Richardson et al. 2007; Cantor et al. 2010). Although none of the DBPs induced bladder cancer in rodents, none were ever tested via the inhalation or dermal route, which may be the critical route-of-exposure, as discussed later. Tumor-site concordance between rodents and humans is complex, but when the route-of-exposure is similar, there is frequently a reasonable correlation (Baan et al. 2019).

Genotoxicity of DBPs

The four THMs were the only DBPs that had been tested for mutagenicity (in Salmonella) by 1979, with the three brominated forms being mutagenic and chloroform being negative (Simmon et al. 1977). Today, 32 DBPs have been evaluated for mutagenicity in Salmonella or other mutagenicity assays (Cortés and Marcos 2018; Richardson et al. 2007; Wagner et al. 2012), and >100 DBPs have been evaluated for DNA damage via the comet assay in Chinese hamster ovary (CHO) cells, including haloacetic acids, haloacids, halophenolics, haloacetonitiriles, nitrosamines and nitramines, cyanogen halides, halomethanes, haloacetaldehydes, halonitromethanes, and haloacetamides(Wagner and Plewa 2017). Nearly all DBPs evaluated are genotoxic, i.e., cause DNA damage and/or mutagenicity, and most are direct-acting agents whose genotoxicity is reduced by the addition of mammalian metabolic activation (S9).

Detailed summaries of the genotoxicity of DBPs have been published (Cortés and Marcos 2018; Richardson et al. 2007; Plewa et al. 2017; Wagner and Plewa 2017), and several of the main conclusions are noted here. With few exceptions, the genotoxic and cytotoxic potencies of the DBPs rank as follows: iodinated > brominated > chlorinated. With the exception of many of the nitrosamines (Wagner and Plewa 2017), nitrogen containing DBPs are generally more cytotoxic and genotoxic both in CHO cells (Plewa et al. 2017) and Salmonella (Kundu et al. 2004) than their carbonaceous homologues. Due to the different endpoints of the assays, the DBPs generally rank differently based on genotoxic potency in CHO cells and Salmonella (Kundu et al. 2004); however, the nitrosamines rank similarly between the two assays (Wagner et al. 2012). In addition, the proportion of types of base-substitution mutations induced by DBPs varies, with some inducing primarily GC to TA and others primarily GC to AT (Figure 1). Most mutations induced by GSTT1-activated brominated THMs are GC to AT base substitutions (DeMarini et al. 1997; Figure 1).

Figure 1.

Data from DeMarini et al. (1995a, 1997) and Kundu et al. (2004). Mutation spectra were determined at the hisG46 base-substitution allele in Salmonella TA100, except for those DBPs evaluated in the presence of GSTT1–1, where strain RSJ100 was used. RSJ100 expresses the rat GSTT1–1 gene and is a derivative of strain TA1535, which is the parent strain of TA100. All data were generated in the absence of metabolic activation (S9 mix). OZ H₂O, ozonated water; MX, mutagen X [3-chloro-4-(dichloromethyl)-5-hydroxy-2(5H)-furanone]; Cl H₂O, chlorinated water; CHN H₂O, chloraminated water; DBM, dibromomethane; BCM, bromochloromethane; at BCNM, bromochloronitromethane; DCA, dichloroacetic acid; CNM, chloronitromethane; TBM, tribromomethane; BDCM, bromodichloromethane; and DBCM, dibromochloromethane.

As assessed by the comet assay in CHO cells, many DBPs, especially the haloacetonitriles and haloacetamides, are more cytotoxic and genotoxic than the 11 DBPs regulated by the U.S. EPA (Wagner and Plewa 2017). The brominated THMs are enzymatically activated to mutagens by glutathione-S-transferase theta-1 (GSTT1), as demonstrated in a transgenic strain of Salmonella that expresses the rat GSTT1 gene (Pegram et al. 1997). However, no mammalian cell lines (including CHO cells) express this gene in culture (Landi et al. 2003), preventing the determination of the genotoxic potency of the GSTT1-activated forms of the brominated THMs in mammalian cells.

Genotoxicity of Drinking Water and Source Water

There were six studies on the mutagenicity of organic extracts of municipal drinking water by 1979 (Loper 1980); today there are >400 studies on the genotoxicity of drinking water extracts based on a recent literature search (DeMarini, unpublished observation) and as summarized by Cortés and Marcos (2018) and Richardson et al. (2007). Although most samples of drinking water are mutagenic, their potencies in the Salmonella mutagenicity assay vary by nearly an order of magnitude, indicative of the variety of organics and the concentrations of those organics in the source waters and the various treatment methods used.

In general, ozonated waters are less genotoxic than chloraminated waters, which are less genotoxic than chlorinated waters both in Salmonella and CHO cells (DeMarini et al. 1995; Daiber et al. 2016; Ersan et al. 2019; Yang et al. 2014). Ground waters (i.e., waters from underground, such as well water from aquafers) used as source waters typically have lower concentrations of organics than surface waters (i.e., waters from surface bodies of water such as rivers or lakes) and, depending on the type of disinfection methods used, drinking water that uses ground water as its source may, therefore, have lower mutagenic potencies than drinking water that uses surface waters as its source (Daiber et al. 2016; Postigo et al. 2018; Takanashi et al. 2009; Warren et al. 2015). The majority (64–96%) of the mutations induced by organic extracts of chlorinated, ozonated, or chloraminated water in strain TA100 of Salmonella are GC to TA base substitutions (DeMarini et al. 1995a; Figure 1).

As noted earlier, carcinogens and carcinogenic activity had been found in surface waters prior to 1979; however, no surface waters or ground waters used as source waters for drinking water had been evaluated for mutagenicity before that time. Researchers have evaluated 104 sites multiple times over a 20-year period in the State of São Paulo, Brazil, some of which are source waters for drinking water, and 20% of the samples were mutagenic (Roubicek et al. 2020).

A review of the literature by Ohe et al. (2004) found that 15% of surface waters were mutagenic, and 3–5% were extremely mutagenic, which was defined as >5,000 revertants/liter-equivalent of water based on analyses using the Salmonella (Ames) mutagenicity assay. The surface waters reviewed by Ohe et al. (2004) were not used necessarily for drinking water and had been selected for mutagenicity analysis based on criteria established by each researcher. Among 36 surface waters in the U.S., some of which are used as source waters for drinking water, 25% were mutagenic, and all of these had low mutagenicity (Berninger et al. 2019), which is defined by Ohe et al. (2004) as <500 revertants/liter-equivalent of water. Collectively, these studies indicate that most surface waters studied to date are either not mutagenic or generally weakly mutagenic.

Genotoxicity of Swimming Pool and Hot Tub (Spa) Water

Although not considered at the time of the first DBP regulation in 1979, swimming pool and hot tub (spa) waters have become important areas of study because they can represent peak DBP exposures. The first report that swimming pool water was mutagenic was by Honer et al. (1980) in Vancouver, BC, Canada. Now there are 22 papers describing the genotoxicity of swimming pool and hot tub (spa) waters; most of this literature is summarized in recent papers and involve data derived primarily from comet assays for DNA damage in cultured mammalian cells and mutagenicity in Salmonella (Daiber et al. 2016; Richardson et al. 2010; Richardson and Ternes 2018).

Studies have identified >100 DBPs in pool and hot tub waters, some not found in drinking water (Daiber et al. 2016; Font-Ribera et al. 2019; Richardson et al. 2010; Richardson and Ternes 2018; Zweiner et al. 2007), especially nitrogen containing DBPs, which are formed from the presence of uric acid from urine in these waters (Richardson et al. 2010). Unlike drinking water, which is not disinfected by bromination, pool and hot tub waters sometimes are, and the concentrations of DBPs and levels of mutagenicity are generally higher in brominated versus chlorinated pools and hot tubs (Richardson et al. 2010; Daiber et al. 2016). In addition, the toxicity of swimming pool water with bromide ions may be as much as 30 times higher than of pool water not containing bromide ions (Hansen et al. 2012).

Other types of disinfection methods, such as ozone and UV light, and different types of waters, such as groundwater or seawater, influence the DBP composition, concentration, and genotoxicity (Richardson and Ternes 2018). Human inputs (sweat, urine, cosmetics, etc.) also contribute to the increase in DBP concentrations and mutagenicity of pool and hot tub water relative to drinking water, with some hot tubs after extensive use having mutagenic potencies >8 times higher than the tap water used initially to fill them (Daiber et al. 2016).

Molecular Epidemiology of Swimmers

In a swimmer study in which the concentration in the pool water of the total THMs was 45.4 μg/L and that of the brominated THMs was 28.3 μg/L, the 49 swimmers on average had significantly increased levels of urinary mutagenicity and increased frequencies of micronuclei (in peripheral lymphocytes) after swimming 40 min (Font-Ribera et al. 2010; Kogevinas et al. 2010). The increases in both of these genotoxic endpoints in the swimmers were associated with increased concentrations of bromoform exhaled in the breath of the swimmers. After swimming, the total concentration of the THM4 was 7 times higher, the frequency of micronuclei in the peripheral blood lymphocytes was 1.2 times higher, and the mutagenic potency of the urine was 2 times higher compared to before swimming.

A similar study done several years later in the same pool with water having similar THM4 concentrations but now with double the concentration of chloroform and only one-third the concentration of brominated THMs found no genotoxicity after 100 swimmers swam for 40 min (Font-Ribera et al. 2019). The data are summarized in Table II and illustrate the important role of brominated THM exposures on the genotoxicity of swimming pool water to swimmers. The concentration of non-mutagenic chloroform is unrelated to genotoxicity in the swimmers, whereas the concentration of the mutagenic brominated THMs is consistent with these genotoxicity data.

TABLE II.

Genotoxicity of Swimming Relative to THM Concentrations in the Pool Water and in Exhaled Breath

Concn in Pool Water (μg/L)			Concn in Exhaled Breath (μg/m3)			Genotoxicity
Total THM	Chloroform	Br-THMs (%)a	Total THM	Chloroform	Br-THMs (%)a	Lymphocyte micronuclei	Urinary mutagenicity	References
45.4	16.1	28.3	7.9	4.5	3.5	+	+	Kogevinas et al. (2010) Font-Ribera et al. (2010)
48.5	37.3	9.5 (66%)	14.4	11.5	2.7 (23%)	−	−	Font-Ribera et al. (2019)

^aPercent decline in the concentration of the Br-THMs in the 2019 study relative to the two 2010 studies.

There are only two bladder cancer case-control epidemiology studies that have evaluated the amount of THM exposure and risk for bladder cancer associated with swimming. Swimmers had an increased risk for bladder cancer in a study in Spain (Villanueva et al. 2007); however, swimmers in New England who had lower exposures to THMs, including brominated THMs, had no added risk for bladder cancer associated with swimming (Beane-Freeman et al. 2017). The dose-response function of the New England study was consistent with that of the Spanish study, although it overlapped across only the lower levels of exposure covered by the New England study.

Molecular Epidemiology of Bladder Cancer and Drinking Water

Until 2007, most epidemiological studies found low relative risks for cancer, primarily bladder and colorectal cancer, associated with exposure from THMs or chlorinated drinking water (IARC 1991, 2004; Villanueva et al. 2003; Villanueva et al. 2004). This changed when Villanueva et al. (2007) for the first time assessed route of exposure and found that the dermal/inhalation route elevated cancer risk more than the oral route. Later analyses confirmed the validity of the exposure assessment (Salas et al. 2013).

Starting in the 1990s, laboratory researchers began mechanistic studies on the brominated THMs because the brominated THMs had greater carcinogenic potency in rodents relative to chloroform. In addition, some of the brominated THMs induced intestinal tumors in rodents (Richardson et al. 2007). Although some studies had reported associations of colorectal cancer with drinking water exposures in humans by that time (IARC 2004), a more recent analysis does not support such an association (Villanueva et al. 2017). As discussed below and in the following section, molecular epidemiological assessment of the Villanueva et al. (2007) study by Cantor et al. (2010) provided mechanistic support for the brominated THMs and the haloacetic acids and the influence of genotype on bladder cancer risk.

In a case-control bladder cancer study in Spain with subjects who had received relatively high exposures to THMs, including brominated THMs, there was a striking increase in bladder cancer risk (odds ratio = 5.9, with CI of 1.8–19.0 among the top quartile of exposure) associated with the combined at-risk genotype of GSTT1, which metabolizes the brominated THMs to mutagens, and a single-nucleotide polymorphism (SNP) in GSTZ1 (GSTZ1 rs1046428 CT/TT), which results in less inactivation of haloacetic acids (Cantor et al. 2010). These two susceptible genotypes occur together in ~24% of the U.S. population (Regli et al. 2015). People without GSTT1 had no significant DBP-associated increased risk for bladder cancer, regardless of the level of DBP exposure, whereas people with GSTT1 did (Cantor et al. 2010). Although these data do not provide information on the possible role of other classes of DBPs and risk for cancer in humans, they do not rule out a role for classes other than the brominated THMs and the haloacetic acids.

A Hypothesis for Induction of Bladder Cancer in Humans by Drinking Water

The two genotoxicity studies in swimmers and the two bladder cancer case-control studies discussed above involving exposures to high versus low levels of THMs and/or brominated THMs, along with pharmacokinetic modeling (Kenyon et al. 2016a,b; 2019) support a hypothesis for drinking-water associated bladder cancer illustrated in Figure 2. The hypothesis is based on the demonstrated higher blood concentrations in humans of brominated THMs via dermal/inhalation exposure compared to oral exposure (Leavens et al. 2007). After dermal/inhalation exposure, the brominated THMs circulate systemically, initially bypassing the liver, and then they are activated to mutagens in the bladder, which may express GSTT1 at higher levels than other tissues. In contrast, oral exposure to these DBPs would first go to the liver where they would be detoxified by CYP2E1.

Figure 2.

A hypothesis for the induction of bladder cancer from primarily the brominated THMs and haloacetic acids in drinking water (Cantor et al. 2010; Regli et al. 2015). No mutations would be induced via the oral route if the DBPs were inactivated by CYP2E1. GC to AT mutations are the predominant base-substitution mutation found in bladder tumors of non-smokers (Alexandrov et al. 2016).

The haloacetic acids are much less volatile and less able to be inhaled and adsorbed through the skin than are the THMs, so oral exposure to the haloacetic acids by individuals containing a specific single-nucleotide polymorphism (SNP) in GSTZ1 (Cantor et al. 2010) would result in less-efficient detoxification of the haloacetic acids. Mono-haloacetic acids can form reactive oxygen species, which are mutagenic and could contribute to the carcinogenicity of drinking water (Pals et al. 2013). In addition, various haloacetic acids can alter mitochondrial metabolism, which also might contribute to cancer risk in humans (Dad et al. 2018).

Additional support for the hypothesis in Figure 2 derives from the finding that GC to AT mutations, which are the primary base-substitution mutation induced by the GSTT1-activated brominated THMs in Salmonella (Figure 1), are also the predominant base-substitution mutation found in bladder tumors of non-smokers (Alexandrov et al. 2016), some of whose cancer may be associated with brominated THM exposure from drinking water. In contrast, GC to TA mutations are the primary mutation induced by tobacco smoke (DeMarini et al. 1995b) and the primary mutation in bladder tumors of smokers; smoking is associated with most bladder tumors (Alexandrov et al. 2016). There is also evidence that DBP exposures can alter DNA methylation levels both in rodents and humans (Salas et al. 2015), supporting a role for epigenetic changes, which are known to play a role in cancer.

Recent Regulations of DBPs and Reduced Occurrence of DBPs

After the U.S. EPA issued its DBP regulation in 1979, other countries began doing so, such that now most European countries, along with Canada, China, Japan, Australia, South Africa, and others regulate or have guidelines for 5 (China) to 11 (U.S.) DBPs (Cortés and Marcos 2018; Evlampidou et al. 2020; Richardson and Ternes 2018). All these countries regulate or have guidelines for all four THMs, some with maximum contaminant levels (MCLs) as low as 25 μg/L. However, the U.S. regulates more DBPs than any other country: the four THMs, five haloacetic acids, bromate, and chlorite (Cortés and Marcos 2018; U.S. EPA 2019). The evolution of these regulations has been chronicled (Richardson et al. 2007; Regli et al. 2015), and new regulations and regulatory methods have been reviewed (Richardson and Ternes 2018).

A recent analysis by Seidel et al. (2017) showed that since the implementation of the Stage 2 D/DBP Rule (U.S.EPA 2006), there has been a reduction of 20 μg/L in the average total THM occurrence for the highest 5% of systems serving >100,000 people across 44 U.S. states in 2010–2014 compared to 1997–1998. Their analysis also found that after implementation of the Stage 2 Rule, the average total THM concentration in large water systems in the U.S. was 30.5 μg/L.

Recently, Evalampidou et al. (2020) estimated that 4.9% of the bladder cancer cases among the population of the European Union are due to THM exposure via drinking water, accounting for 6,561 bladder cancer cases per year. The mean THM concentration was 11.7 μg/L, and the population-attributable fraction ranged from 23.2% for Cyprus to 0% for Denmark and the Netherlands, which do not chlorinate their water. Thus, even with improved water-treatment methods and regulatory measures, some small proportion of bladder cancer in countries that chlorinate their water appears attributable to DBPs.

ADDITIONAL RESEARCH NEEDS ON DRINKING WATER AND CANCER

Further improvements in analytical methods, water-treatment procedures, and toxicological and epidemiological assessments will provide guidance for future DBP regulations. Many emerging issues and research needs have been identified (Dong et al. 2019; Li and Mitch 2018; Regli et al. 2015; Richardson and Plewa 2020; Richardson et al. 2007), including the (a) characterization of the remaining total organic halogen (TOX) in some chlorinated waters that is still of unknown chemical composition; (b) greater use of analytical methods beyond gas chromatography-mass spectrometry (GC-MS) for identification of additional DBPs (Yang et al. 2019), such as liquid chromatography-mass spectrometry (LC-MS) and Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR MS) (H.K. Liberatore et al., in prep.), utilizing both electrospray ionization (ESI) and other ionization modes, such as atmospheric pressure chemical ionization (APCI); (c) application of disinfectant-reaction chemistry to precursors of known structure (Li and Mitch 2018); (d) use of high-throughput computational toxicology and other effects-based bioassays, coupled with bioassay-directed fractionation to characterize the toxicity of waters (Berninger et al. 2019); (e) application of computational methods for exposure assessment (Warth et al. 2017); (f) further studies on the toxicology and formation of the highly toxic iodo-DBPs, (g) use of additional biomarkers coupled with improved exposure assessments beyond THMs in molecular epidemiology studies (Kimura et al. 2019); (h) systematic assessment of various water-treatment methods alone and in combination with a variety of source waters (Cuthbertson et al. 2019); (i) application of additional molecular methods to determine further mechanisms by which DBPs cause cancer; and (j) better integration of basic and applied research with regulatory science.

Improved toxicological assays are needed in vivo and in mammalian cells in vitro that express the GSTT1 gene, permitting the detection of the brominated THMs relative to other DBPs within the same bioassay. Nearly all of the focus for the past 40 years has been on genotoxicity in vitro—either mutagenicity in Salmonella or DNA damage in CHO cells. Assays that permitted both to be determined in the same mammalian cell line or rodent would permit comparison between the two endpoints in the same system. Other endpoints associated with cancer, such as oxidative stress and other types of DNA damage and mutation, as well as alterations in gene expression, should be incorporated into a single cell line or rodent assay. The development of appropriate computational toxicological assays, especially those detecting mutation and DNA damage, would permit a high-throughput assessment of large numbers of water extracts for genotoxicity. Additional molecular epidemiology studies of swimmers seem warranted given the limited but suggestive studies of the genotoxicity and potential carcinogenicity of chlorinated water via dermal/inhalation exposure that can be studied uniquely in this population.

CONCLUSIONS AND A PERSPECTIVE

Starting with limited knowledge prior to the 1970s of (a) the existence of just four THMs and their mutagenicity, (b) the rodent carcinogenicity of just two of the four, and (c) 19 epidemiology studies showing an association between source water or drinking water and risk for cancer, scientists worldwide began a cooperative effort to understand whether drinking water caused cancer, which was a critical scientific issue at the time.

After more than 40 years of work, >600 DBPs have been identified, the mechanisms by which they are formed and the concentrations at which they occur are reasonably well understood, the genotoxicity and general toxicity of >100 DBPs and thousands of water samples have been characterized, 22 DBPs (not counting numerous nitrosamines) have been evaluated for cancer in rodents, and a variety of treatment and disinfection methods have been developed and are being applied to reduce the concentration of DBPs while providing microbiologically safe drinking water.

Most developed countries use modern water-treatment methods and have DBP regulations or guidelines, and the U.S. EPA has reduced the allowable concentration of the combined four THMs from 100 to 80 μg/L, along with regulating seven additional DBPs. Thus, the potential risk for bladder cancer today from drinking water in many regions around the world is likely reduced compared to what it was prior to this research effort. This is a testament to the dedicated, cooperative efforts of thousands of scientists, with support from the public and private sectors, to provide techniques and knowledge to regulators to improve public health.

At least one hypothesis has been developed from mechanistic work in vitro, in rodents, and humans to explain how drinking water might cause cancer in a small but significant population group that experiences both a sufficient exposure and possesses an at-risk genotype (Cantor et al. 2010). Further studies from basic research to molecular epidemiology are needed to confirm this model and identify any additional models that would account for the cancers associated with exposure to drinking water.

Although much has been learned during the past 40 years, there are populations worldwide that are not benefitting from advanced water-treatment methods and improved regulations, and/or are using impaired source waters, resulting in less-than-desired levels of DBPs and other contaminants (Furst et al. 2019). In addition, studies have identified health effects other than cancer that are associated with drinking water, especially reproductive and developmental effects (Grellier et al. 2010; Kaufman et al. 2018; Legay et al. 2010; Smith et al. 2016; Wright et al. 2017).

Perhaps more pressing than the relatively small cancer risk posed by DBPs is the lack of microbiologically safe drinking water for a significant proportion of the population worldwide and the reduction in source water suitable for drinking water (Hutton and Chase 2017; Maddocks et al. 2015; Shaffer et al. 2019). The potential increased use of impaired water sources in for drinking water could result in an increase in DBP concentrations in the absence of strict regulatory controls on source water and treatment methods. These new challenges will require as much innovative research and government and private-sector engagement as that initiated 40 years ago to address the issue of DBPs and the risk for cancer.