A lack of regulation is due in part to uncertainty over what cyanotoxin concentration and duration of exposure is considered safe. As a first step, the United States Environmental Protection Agency (USEPA) placed cyanobacteria biomass (draft 1) or cyanotoxins (anatoxin-a, MCLR, and cylindrospermopsin) on their Contaminant Candidate Lists since 1998 (draft 1 and 2) and 2009 (draft 3), respectively. Draft 4 to be released soon is limited to these three cyanotoxins (e.g., saxitoxin is not listed). In some cases, it may be possible to derive numerical limits based on the most current literature. Below we discuss numerical limits for the most common cyanotoxins in the Great Lakes region for which there is sufficient toxicological data.
The majority of countries with some form of regulation or established health advisories follow closely to the WHO provisional guideline value of 1 µg/L/day MC in drinking water. This limit is based on the tolerable daily intake level of 40 µg/kg·b.w. per day determined by Fawell et al. [155
] and Falconer et al. [156
], a mean body weight of 60 kg, and a 1000 fold uncertainty factor. The uncertainty factor includes a factor of 100 for intra-and interspecies variation in toxicity as well as a factor of 10 for lack of information on MC toxicity, particularly chronic exposures. It was also assumed adults drink 2 L of water per day and that 80% of this is from tap water. Thus, this guideline is primarily directed at adults.
Children as opposed to adults have most often been the victims in human poisoning events [295
] and are more at risk for illness as a result of cyanotoxin exposure given a relatively smaller body size proportional to the volume of water they consume and the potential for developmental toxicity [297
]. Weirich and Miller [297
] recalculated the WHO guideline value adjusting the mean body weight to that of children aged 9 months, 5.5, or 9.5 years. In addition, daily drinking water intake was adjusted to 1 L for all ages. Based on this calculation for children aged 5.5 years and 9 months the guideline value was reduced to 0.6 and 0.3 µg/L, respectively.
The USEPA recently published a health advisory for MCs in drinking water [301
]. They established provisional guideline values for two different age groups, (1) 0.3 µg/L for bottle-fed infants and pre-school aged children under 6 years of age, and (2) 1.6 µg/L for children older than 6 years of age and adults. Both are ten-day guideline values, meaning that exposure for ten days below these levels is considered protective of public health, but effects of exposure at any level for longer than 10 days is unknown. The guideline values were calculated by dividing a lowest-observed-adverse effect level (LOAEL) of 50 µg/kg·b.w./day by the product of an uncertainty factor (1000) multiplied by the drinking water intake rate (L/day) normalized to mean body weight of each age group.
Reference dosages such as the LOAEL/NOAEL are typically taken from animal exposure studies. While a number of animal studies have described the toxicity of MCs, relatively few studies have quantified toxicity of MC in a rodent model using pure compound and in well-described, controlled repeat oral dose trials at multiple exposure levels. Oral exposure compared to other exposure routes (e.g., IP injection) in animal models, mimics human exposure to cyanotoxins in drinking water, as well as ingestion recreationally. In addition, use of pure compound as opposed to mixtures or crude cell extracts is necessary in order to establish health effects of each compound independently, though research is needed on the effects of defined cyanotoxin mixtures.
A NOAEL of 40 µg/kg·b.w./day derived from Fawell et al. [155
] and supported by a separate study by Falconer et al. [156
] was used by the WHO in calculating their provisional guideline level. The LOAEL for calculating the EPA guideline value for MCs was based on the study by R. Heinze in 1999 [157
]. The USEPA advisory for MCs used this study to determine that the LOAEL for MCLR is 50 µg/kg·b.w./day via the oral route with the major end point being liver lesions. The study by Heinze did not test lower dosages to determine if they would also cause liver damage, so based on that study alone it is not known if 50 µg/kg·b.w./day is the lowest dose causing liver lesions. Modeling the dose-response relationship might provide evidence that lower doses would produce no or minimal adverse effects, but with only two exposure groups it would be difficult to establish a statistically significant dose-response curve. As such, the USEPA applied an uncertainty factor of 3 to account for extrapolating from a LOAEL to a NOAEL, meaning that the NOAEL could be as low as 16.7 µg/kg·b.w./day. This may not be sufficient for other adverse effects reported for MCs including reproductive and neurotoxic effects as described above.
A LOAEL of 50 µg/kg·b.w./day is at least partially corroborated by the Fawell et al. study, which found no significant pathological changes in the liver at 40 µg/kg·b.w./day MCLR. Yet, the Fawell and Heinze studies are not directly comparable because the Fawell et al. study used mice, not rats, in their 13-week repeat oral dose study. In addition, mice were given MCLR by oral gavage whereas the Heinze study provided five rats per cage with 150 mL/day of drinking water containing MCLR, 93–97% of which was consumed by the animals. The Heinze study was also for a shorter duration (28 days), which might be more realistic for calculating a 10-day health advisory.
The EPA guideline value takes into account the drinking water intake rate by mean body weight ratio (DWI/BW) of each age group. These values were derived from existing data. For children <3 months old the data used by EPA were from the USDA Continuing Survey of Food Intakes by Individuals program 1994–1996, and 1998 [302
], whereas data for all other age groups were from the National Health and Nutrition Examination Survey 2003–2006 [303
]. Young children as opposed to teens and adults have a higher daily water intake rate per kg of body weight. For example, infants <3 months old have a DWI/BW over ten times that of 16 to 18 year olds (Figure 13
). Unless there is some reason to expect that infants and children will not be consuming municipal drinking water, the guideline values for cyanotoxin exposure need to be directed at the most sensitive age groups.
The USEPA advisory uses DWI/BW data at the 90th percentile of usage to calculate the guideline values. The resulting values for various age groups then ranges from 0.2 µg/L for infants to 2.2 µg/L for 3–6 year olds at this 90th percentile level. A guideline value of 0.3 µg/L was chosen instead of 0.2 µg/L for children <6 years of age. The EPA advisory states that the uncertainty factor of 10 for intraspecies variation adequately accounted for this difference and reflects the average DWI/BW ratio of children under 6. That uncertainty factor included 10 for inter-and 10 for intraspecies variability, 3 for lack of data, and 3 for extrapolating from a LOAEL to NOAEL.
The percentile category of DWI/BW data used affects the calculated guideline value considerably. For example, for children under 3 months of age, arguably the most sensitive population, the calculated guideline value ranges from 0.16 to 4.34 µg/L depending on if DWI/BW data are taken from the 99th or 10th percentile respectively (Figure 13
). Thus, using the EPA formula, the most conservative guideline value would be 0.16 µg/L for bottle-fed infants.
There is some question as to whether a provisional guideline value can be applied to children and infants when the LOAEL used to calculate that guideline value was derived from a study of adult male rats. There is some evidence in the literature for potential developmental effects of MCs [304
]. As such it is possible that individuals of a younger age could develop adverse effects later in life due to an exposure at levels lower than 50 µg/kg·b.w./day, during critical periods of development.
Liver damage may not be the only adverse health effect end point to consider when developing numerical limits for MCs. While the liver is clearly one of the most affected organs, MCs also affect brain and reproductive tissues [158
The study by Li et al. reported central nervous system toxicity after repeat oral dose exposures of MCLR in rats. The EPA health advisory does not consider this study in its formulation of the drinking water guideline value for MCs or neurotoxicity of MCLR as an end point. One problem with the Li et al. study is that the MCLR stock (1 mg) was dissolved in 1 mL methanol to a concentration of 1000 µg/mL, then diluted ten-fold with pure water to a working stock concentration of 100 µg/mL. At that point the working stock was 10% methanol. This working stock was then diluted with water to produce drinking water at 0.2, 1.0, and 5.0 µg/mL or 0.02%, 0.1%, and 0.5% methanol. Thus in addition to varying MCLR dosage, rats received varying levels of methanol, a known neurotoxin [306
]. By comparison, the Heinze study dissolved 20 mg MCLR in absolute ethanol (10,000 µg/mL) to make a stock solution, this was diluted in pure water to 1000 µg/mL working stock (10% ethanol) and diluted further by an unknown amount to produce 150 mL of MCLR laden drinking water for 5 rats. The use of ethanol over methanol as a solvent is beneficial as ethanol is much less toxic than methanol [308
Li et al. [170
] also examined the developmental effects of MCLR in a repeat maternal oral dose study in rats showing deficiencies in the Morris water maze test of offspring due to MCLR exposure in utero. As with Li et al. (2014), the study by Li et al. (2015) used methanol to make the stock solution of MCLR resulting in trace levels of methanol in all exposures. However, in the Li et al. (2015) study methanol was included in the negative control and was normalized to 0.002% (i.e., 1.6 mg/kg·b.w.) in all treatments including the control. The EPA health advisory for MCs does not consider this study in calculating the guideline value for MCs in drinking water. The EPA health effects support document [298
] indicates that the study is confounded because there may be synergy between methanol exposure and MCLR.
Reproductive toxicity is another endpoint to be considered in formulating guideline values. Chen et al. [171
] show effects of MCLR on male reproductive tissues (e.g., sperm motility/counts, testis weight) in repeat dose oral exposure studies in mice. The EPA health advisory considered this study in deciding whether to include reproductive effects as an endpoint, but determined that the study design was lacking in several ways. The EPA indicates that no testis weights were given, no methods were given for how sperm was handled, or how percent sperm motility was determined, the purity of MCLR, species and ages of mice, body weight, amount of water consumed, and dosage levels.
Chen et al. indicate that sperm counts and motility were determined using the HTM-TOX IVOS semen analyzer and that MCLR was purchased from Alexis Biochemicals (Enzo Life Sciences), which sells MCLR at >95% purity. The species of mouse is not given, but their body weights are given (15–25 g) which would allow one to estimate the exposure level. Water intake across 28 mouse strains ranges from 4 to 8 mL per day [309
]. The lowest level at which effects were observed was at the 3.2 µg/L level. Therefore, at this exposure level the dosages were likely between 0.9 and 1.7 µg/kg·b.w./day for a 15 g mouse consuming between 4 and 8 mL per day, or between 0.5 and 1.0 µg/kg·b.w./day for a 25 g mouse consuming between 4 and 8 mL per day, an overall mean estimated dosage of 1.0 µg/kg·b.w./day at the 3.2 µg/L level. The EPA Health Support Document for MCs calculates a LOAEL and NOAEL of 0.79 and 0.25 µg/kg·b.w./day for reproductive toxicity based on this study. This suggests that adverse effects caused by MCLR on reproductive tissues occurs at doses well below the LOAEL for liver toxicity of 50 µg/kg·b.w./day.
If the guideline values were calculated for endpoints of reproductive and neurotoxicity using the aforementioned studies then they would be lower than guideline values based on liver toxicity. Using the Chen et al. [171
] study for a reproductive toxicity LOAEL (1.0 µg/kg·b.w./day), and the Li et al. study for a neurotoxicity LOAEL (5 µg/kg·b.w./day) the guideline value at the 90th percentile DWI/BW ratio ranges from 0.004 to 0.029 µg/L/day for reproductive toxicity and 0.02–0.15 µg/L/day for neurotoxicity across age groups from <3 months to adults. The mean guideline values for children <6 years old would be 0.01 µg/L and 0.07 µg/L for reproductive and neurotoxicity, respectively (Table 4
). These are 4- and 30-fold lower than the EPA guideline value for liver toxicity in children <6 years old. Given these large differences it is critical that future studies confirm the level of MC that displays reproductive and neurotoxicity in repeat oral dose study designs.
EPA and WHO guideline values are all based on studies of the toxicity of the MCLR congener. However, MC congers vary widely in their potency from those that are non-toxic to those that are as toxic or more than MCLR based on in vitro assays or animal studies [310
]. As discussed above the range of potency displayed by MC congeners is primarily due to (1) recognition and transport rates of different MC congers by OATP receptors and (2) presence/absence of an intact MDHA moiety at position 7 in different MC congeners affecting binding to cysteine residues in phosphatases [314
]. Thus, guideline values based only on studies of MCLR toxicity may not reflect the true toxicity of mixtures of MC variants that occur in nature.
Monitoring and analytical methods for MC detection will necessarily be influenced by the drinking water guideline value. The EPA advisory suggests MCLR is a suitable surrogate for the toxicity of all MCs because it is one of the most commonly occurring or monitored in the environment and is one of the most toxic congeners. Furthermore, the EPA advisory indicates that the guideline values apply to total MCs in a given sample. MCs can exist bound to proteins or other molecules with thiol groups (e.g., free cysteine), transformed, or otherwise in a non-toxic state. The guideline value should be applied to the total toxic MCs (TTMC), which would include non-transformed, non-protein bound total MCs capable of covalently binding to and inhibiting protein phosphatases. Methods for specifically quantitating the TTMC pool in drinking water have not been demonstrated. In theory, the MC ELISA would measure non-toxic protein bound as well as toxic, non-protein bound MCs. The EPA advisory suggests using EPA Method 544 for quantification of intracellular and extracellular MCs using LC-MS/MS in drinking water. However, this method does not target TTMCs and the surrogate standard indicated for use in that method is no longer available. The PP1/2A inhibition assay might be useful as an activity assay in quantifying TTMCs, but it would also detect other PP1/2A inhibitors mentioned above. Further complicating this issue is the fact that a recent study shows MCs covalently bound to thiols can deconjugate over time releasing a fully intact MC molecule. If the guideline value is for TTMCs then method development for detecting these compounds is critical.
According to Munday and Reeve [318
] currently no sub-acute repeat oral dosing studies of saxitoxin in animals have been reported using approved protocols. In contrast, the LD50
for saxitoxin via the oral route is well known at approximately 200 µg/kg·b.w. in the mouse (Wiberg and Stevenson) and ranges from 91 µg/kg·b.w. in pigeons to 800 µg/kg·b.w. in monkeys (Table 2 in [319
A LOAEL for saxitoxins of 1.5 µg/kg·b.w. has been reported by the European CONTAM on marine biotoxins based on epidemiological data [290
]. Unlike other cyanotoxins, human poisonings with saxitoxins or PSPs have been well documented due to frequent human consumption of shellfish containing hazardous levels of saxitoxins [320
]. CONTAM reviewed approximately 500 reports of human illness and death from eating shellfish contaminated with saxitoxins. In all cases concentrations of saxitoxins in the shellfish consumed were determined by the mouse bioassay. As such, the congeners of saxitoxins consumed are unknown and data are provided in mouse units. Mouse units were converted to mass per volume of saxitoxin equivalents by multiplying the mouse units by 0.18 µg saxitoxin equivalents per kg, which is widely used as conversion factor [323
The use of case reports of human illness often requires a number of assumptions in order to relate dosage to the human health outcome. The reports of human intoxication events used by CONTAM required several such assumptions. For example, in many cases the weight of victims was unknown and assumed to be 60 kg if an adult or average age adjusted weight [324
]. In one case, the amount of shellfish consumed was estimated based on the number of empty shells found after the meal [324
]. In another case, the amount of toxin present in the shellfish consumed was estimated by interpolating concentration in shellfish collected the day before and after the poisoning event [325
]. In addition, details of how the mouse bioassay was performed were often not available making it unclear if the assay was performed in agreement with the FDA approved method, 959.08 from the Association of Official Analytical Chemists.
These assumptions contribute uncertainty in the calculation of the LOAEL of 1.5 µg/kg·b.w. To account for this, an uncertainty factor of 10 could be used in calculating a guideline value for saxitoxins. On the other hand, since the LOAEL was determined based on human cases of illness an uncertainty multiplier for interspecies differences is not required. In addition, considering a population size of 500 human cases of saxitoxin related illness there is less uncertainty variation in toxicity based on intraspecies differences.
The CONTAM panel on saxitoxins divided the LOAEL by a factor of 3 to derive a NOAEL of 0.5 µg/kg·b.w. A guideline value can be calculated for saxitoxins using the same formula as for MCs, and this LOAEL of 0.5 µg/kg·b.w., as well as the DWI/BW ratio at the 90th percentile, and an uncertainty factor of 10. This produces a calculated guideline value for saxitoxins ranging from 0.2 to 1.5 µg/L across all age group categories (Figure 15
) and 0.3 µg/L using the same ingestion rate applied by the EPA of 0.15 L/day for the first year of life (Table 4
). On the other hand, for infants < 3 months old at the 99th percentile of the DWI/BW ratio, the most sensitive group, the guideline value is 0.16 µg/L.