Perchlorate exposure may be higher in infants compared with older persons, due to diet (infant formula) and body weight versus intake considerations. Our primary objective was to quantitatively assess perchlorate concentrations in commercially available powdered infant formulas (PIFs). Secondary objectives were: (1) to estimate exposure in infants under different dosing scenarios and compare them with the perchlorate reference dose (RfD); (2) estimate the perchlorate concentration in water used for preparing PIFs that would result in a dose exceeding the RfD; and (3) estimate iodine intakes from PIFs. We quantified perchlorate levels in three samples (different lot numbers) of reconstituted PIF (using perchlorate-free water) from commercial brands of PIF in each of the following categories: bovine milk-based with lactose, soy-based, bovine milk-based but lactose-free, and elemental (typically consisting of synthetic amino acids). Exposure modeling was conducted to determine whether the RfD might be exceeded in 48 dosing scenarios that were dependent on age, centile energy intake per unit of body weight, body weight percentile, and PIF perchlorate concentration. We obtained three different samples in each of the five brands of bovine- and soy-based PIF, three different samples in each of the three brands of lactose-free PIF, and three different samples in two brands of elemental PIF. The results were as follows: bovine milk-based with lactose (1.72 μg/l, range: 0.68–5.05); soy-based (0.21 μg/l, range: 0.10–0.44); lactose-free (0.27 μg/l, range: 0.03–0.93); and elemental (0.18 μg/l, range: 0.08–0.4). Bovine milk-based PIFs with lactose had a significantly higher concentration of perchlorate (P<0.05) compared with all. Perchlorate was a contaminant of all commercially available PIFs tested. Bovine milk-based PIFs with lactose had a significantly higher perchlorate concentration perchlorate than soy, lactose-free, and elemental PIFs. The perchlorate RfD may be exceeded when certain bovine milk-based PIFs are ingested and/or when PIFs are reconstituted with perchlorate-contaminated water.
Proper maternal thyroid function during pregnancy as well as adequate infant thyroid function in the post-partum period is vital for optimal neurological development of the child (Haddow et al., 1999; Klein et al., 2001; Grüters et al., 2003; National Academy of Sciences, 2004; van Wassenaer and Kok, 2004; Pemberton et al., 2005). Environmental exposures to agents that have an intrinsic ability to adversely affect the thyroid, such as perchlorate, are an area of concern. The potential impact of these exposures on fetal and childhood development are becoming increasingly important areas of research, especially as our capability to assess exposure through advanced biomonitoring methodologies increases (Needham et al., 2005; Pirkle et al., 2005). Iodine is a necessary component of triiodothyronine and thyroxine, both of which are forms of thyroid hormone (TH) and are necessary for normal growth and development, especially in infants (National Academy of Sciences, 2004). Iodine is usually ingested in the form of an iodide salt that is then transported into the thyroid gland by a specific sodium-iodide symporter [NIS]. It is then oxidized to iodine, which is then used to produce TH (National Academy of Sciences, 2004; Bouchard, 2006). Inadequate dietary intake of iodine or conditions in which uptake of absorbed iodine into the thyroid gland is blocked can lead to an iodine deficiency state. Perchlorate adversely affects thyroid function by competitive inhibition of iodide transport into the thyroid by a sodium-iodide symporter [NIS] on the gland, thereby inducing a functional iodide deficiency state. TH production may be adversely affected in some perchlorate exposure scenarios, especially in persons that are ingesting inadequate amounts of exogenous iodide.
Perchlorate can be found naturally in the environment, but ammonium perchlorate has been used as an oxidizing agent and as an accelerant in the formulation of solid propellants for rockets, missiles, and aerospace systems (such as the space shuttle) since World War II (Soldin et al., 2001; National Academy of Sciences, 2004). It is soluble, and contaminated wastewater has been found in the soil and water table around institutions where perchlorate salts were manufactured, stored, or used (Soldin et al., 2001; Hershman, 2005). In 1985, perchlorate was discovered in wells of Superfund sites in California (National Academy of Sciences, 2004). However, the true extent of perchlorate contamination of water supplies did not begin to be recognized until 1997, when the assay sensitivity for perchlorate in water improved from 0.4 mg/l (parts per million or p.p.m.) to 4 μg/l (parts per billion or ppb) (Soldin et al., 2001; National Academy of Sciences, 2004). Although local perchlorate manufacturing facilities were found to contribute to the problem in California, the primary source of perchlorate contamination was determined to be from water originating in the Colorado River, which was contaminated by a former perchlorate-manufacturing site in Nevada (Soldin et al., 2001).
Since 1985, this problem has extended beyond simple contamination of water supplies used for drinking water. Perchlorate can be absorbed by plant life when in the surrounding soil or water (Yu et al., 2004). It has been detected in conventionally and organically produced leafy vegetables and fruit (Sanchez et al., 2005), bottled water (Krynitsky et al., 2004), and dairy products such as milk (Krynitsky et al., 2004; Capuco et al., 2005; Kirk et al., 2005). The mammary gland has an intrinsic capacity to transport and concentrate iodide, which explains its presence in milk and milk products (Capuco et al., 2005). Perchlorate exposure is ubiquitous in the United States, as established by detectable perchlorate in human urine (Valentin-Blasini et al., 2005; Blount et al., 2006, 2007), and breast milk (Kirk et al., 2007; Pearce et al., 2007). Additionally, perchlorate exposure is associated with changes in thyroid function in US women with low iodine intake (Blount et al., 2006). The recent adoption of a daily perchlorate reference dose (RfD) by the National Academy of Sciences (NAS) and the Environmental Protection Agency (EPA) has facilitated research in this area by establishing this standard for comparison (Greer et al., 2002; National Academy of Sciences, 2004; United States Environmental Protection Agency, 2007). In the NAS report Health Implications of Perchlorate Ingestion (NAS), a recommendation for continued research into potential exposure routes and adverse effects in sensitive populations such as infants is also strongly advised (Greer et al., 2002; National Academy of Sciences, 2004). We are aware of only one earlier report that includes a quantitative analysis of perchlorate in commercially available powdered infant formulas (PIFs) (Pearce et al., 2007). In this study, we sought to expand on these limited results and increase the scope of information on this topic through a systematic collection and analysis of multiple different types of commercially available PIFs. We then discuss these findings, along with the results of other studies, with regard to the recently adopted perchlorate RfD of (0.7 μg/kg per day) (National Academy of Sciences, 2004). We also estimate perchlorate intake in infants from PIF consumption, the perchlorate concentration in water used for preparing PIFs that would result in exposures above the RfD, and iodine intakes from PIFs.
We identified four general types of PIF by their package labeling, including bovine milk-based with lactose, soy-based, bovine milk-based lactose-free (referred to solely as lactose-free for the remainder of this article), and elemental (typically consisting of synthetic amino acids). We then attempted to measure perchlorate in samples of five different brands of PIF within each of these categories. Within each brand, samples with different lot numbers were chosen by convenience. The samples were obtained at five local (to the study investigators workplace) commercial food markets in 2006. Data on each sample with regard to brand, category, and lot number were collected.
Samples of PIF were dissolved in perchlorate-free water (determined by analysis with a method detection limit of 0.05 ng/ml) according to manufacturer instructions and analyzed by use of a modified version of the method of Valentin-Blasini et al. (2005). Briefly, 18O4-perchlorate internal standard was added to each sample and vortex mixed. Subsequently, proteins were precipitated by addition of 3 ml of cold ethanol (−20°C). The solution was transferred to a 15-ml conical tube and centrifuged at 3016 g for 35 min at −5°C. The supernatant was removed and dried under a stream of nitrogen gas at 60°C for 30 min. The resulting residue was re-suspended in 1 ml de-ionized water and transferred to a pre-conditioned solid phase extraction cartridge (C18-E, 200 mg/3 ml, Phenomenex). The breakthrough fraction and a subsequent 1 ml water wash were used to elute perchlorate. This solution was vortexed, transferred to an autosampler vial (1 ml glass, Dionex), and analyzed by use of ion chromatography–electrospray ionization–tandem mass spectrometry. Perchlorate was quantified on the basis of the peak area ratio of analyte to stable isotope-labeled internal standard. Two quality control pools were analyzed in each analytical batch with unknown samples. Reported results met the accuracy and precision specifications of the quality control/quality assurance program of the Division of Laboratory Sciences, National Center for Environmental Health of CDC, similar to rules outlined by Westgard (Westgard et al., 1981). We assessed perchlorate contamination by lot screening all reagents and analyzing blanks with each batch of unknowns, and we identified no background perchlorate contamination.
Data were analyzed by use of SAS 9.1. Generalized estimating equations (GEEs) were used to model perchlorate concentrations in the different categories of formula. GEEs also account for the correlation of perchlorate concentrations among samples within brands.
Hypothetical dosing scenarios in infants were created by using geometric mean perchlorate concentrations and the upper or highest value reported for each type of PIF, including bovine milk-based with lactose, soy-based, lactose-free, and elemental. Body weights for these exposure scenarios were based on an average of male and female body weights for the 10th, 50th, and 90th percentiles at the 1- and 6-month-old age groups from established data (US Centers for Disease Control and Prevention, National Center for Health Statistics, 2007). Infant formula ingestion rates were calculated as follows: average (male and female) energy intakes in kcal/kg per day by formula fed infants at the 10th, 50th, and 90th centile energy intakes per unit of body weight at approximately 1 month and 6 months of age (Fomon and Bell, 1993), were multiplied by averaged (male and female), estimated body weights for infants at these ages in these weight percentiles (US Centers for Disease Control and Prevention, National Center for Health Statistics, 2007). The product was then divided by the number of kilocalories (20) per 29.4 ml (one ounce) of formula and the value converted to liters ingested per day. This number was then multiplied by the perchlorate concentration in μg/l and then divided by the average, estimated body weight for that percentile and age to determine estimated daily dose in μg/kg per day. These results were then compared with the perchlorate RfD.
The estimated water perchlorate concentration needed to reach the RfD was calculated by multiplying the perchlorate RfD by estimated body weight (age and body weight percentile dependent) (US Centers for Disease Control and Prevention, National Center for Health Statistics, 2007) and dividing by the estimated ingested daily volume of formula (Fomon and Bell, 1993). The PIF perchlorate concentration was then subtracted from this result.
Total kcal/day values for estimated iodine intakes (Table 1) were calculated based on reported values (Title 21: Food and Drugs, 2008) stratified by age and centile energy intake per unit of body weight (Fomon and Bell, 1993) multiplied by the appropriate value for body weight percentile (US Centers for Disease Control and Prevention, National Center for Health Statistics, 2007). For instance, the 10th percentile value in a 1-month-old infant was calculated by taking the 10th centile energy intake per unit of body weight and multiplying it by the average body weight (males and females) in a 1-month-old infant for the 10th body weight percentile.
We were unable to obtain three different samples (with different lot numbers) of five different brands for each type of PIF after visiting five local stores and deemed it unlikely to find any additional samples at that point. We subsequently ceased sample finding efforts. We were able to obtain three different samples in each of the five different brands of bovine milk-based and soy-based PIF. We were able to obtain only three different samples from each of the three different brands of lactose-free PIFs. Finally, we obtained three samples from each of the two different brands of elemental PIFs. The geometric mean, median, and range of perchlorate concentrations by category are presented in Table 2. Bovine milk-based PIFs with lactose had a significantly higher concentration of perchlorate (P<0.05) than all others. Two widely distributed different brands of bovine milk-based PIFs with lactose had significantly higher perchlorate levels than all of the other bovine milk-based PIFs with lactose (Table 3).
The results for hypothetical dosing scenarios for perchlorate through consumption of PIF are listed in Table 4. One-month-old infants in the 10th, 50th, and 90th percentiles, and 6-month-old infants at the 90th percentile that ingest bovine milk-based PIFs with lactose containing the upper value of perchlorate concentration found on testing are at risk for exceeding the perchlorate RfD. In more than half the cases (26/48), the perchlorate RfD could be exceeded by reconstituting PIF with water that was contaminated with at least 4 μg/l of perchlorate. Estimated daily intake of iodine through PIF, stratified by body weight and age, can be found in Table 1. The results were dependent on estimated quantity of iodine.
Our analysis identified detectable levels of perchlorate in all PIFs that were tested. Conservative modeling of PIF ingestion in infants identified the potential for perchlorate doses in excess of the RfD of 0.7 μg/kg per day in some cases. A brief review of the current perchlorate standard is warranted to achieve a proper comparison.
As the extent of potential exposure routes for perchlorate began to be fully realized from the earlier work discussed, efforts were made to clarify what was known about health effects from perchlorate exposure. As part of this effort, the NAS was commissioned to address this very question as well as to review the key findings of the EPA's 2002 draft perchlorate risk assessment document Perchlorate Environmental Contamination: Toxicological Review and Risk Characterization. In 2004, the NAS released the document Health Implications of Perchlorate Ingestion, which addressed a number of issues. NAS also recommended a perchlorate RfD of 0.0007 mg/kg per day (0.7 μg/kg per day) (Greer et al., 2002; National Academy of Sciences, 2004). The EPA defines the RfD as an estimate, with uncertainty spanning perhaps an order of magnitude, of a daily oral exposure to the human population (including sensitive subgroups) that is likely to be without an appreciable risk of deleterious effects during a lifetime (United States Environmental Protection Agency, 2007). The EPA has also since adopted the NAS recommendation (United States Environmental Protection Agency, 2007). In addition to this recommendation, the NAS also urged that further study on this issue be given to the absence of data on potential adverse effects in sensitive subpopulations, such as pregnant women, their fetuses, and newborns (National Academy of Sciences, 2004).
All bovine milk-based PIFs with lactose had the highest concentrations of perchlorate when those PIFs were compared with the other PIFs tested (Tables 2 and 3). Actual perchlorate testing results do not differ substantially from Pearce et al. (2007), who found perchlorate levels ranging from 0.2–4.1 μg/l in milk-based PIFs (14 samples) and 0.3–0.4 μg/l in soy-based samples (three samples). As mentioned earlier, the mammary gland has an intrinsic capacity to transport and concentrate iodide, much like the thyroid (Capuco et al., 2005). This process occurs through the NIS transporter mechanism, although other active transport mechanisms may also exist (Capuco et al., 2005). As perchlorate is a competitive inhibitor of the NIS transporter, the mammary gland can preferentially transport perchlorate instead of iodide (Capuco et al., 2005; Dohán et al., 2007; Tran et al., 2008). Concentrations of both perchlorate and iodide in dairy products and human milk are influenced by the amount of these items in the diets of animals as well as humans. Human exposure to perchlorate may occur through ingestion of bovine milk products (Capuco et al., 2005) and through human milk by infants (Pearce et al., 2007). This mechanism, which consists of a glandular ability to actively transport and concentrate perchlorate, explains the higher concentration of perchlorate in bovine milk-derived products than in others (e.g., soy-based).
All of the hypothetical dosing scenarios in which the perchlorate RfD (0.7 μg/kg per day) would be exceeded involve consumption of the bovine milk-based PIF with lactose and that have the highest perchlorate concentration (5.05 μg/l). Of additional interest is that affected exposure scenarios included specific scenarios involving all body weight percentiles, not just the lower extreme (10th), which might be expected. This is reassuring at first glance because most scenarios do not exceed the perchlorate RfD; however, these scenarios also assume that reconstitution of PIF is with perchlorate-free water.
Perchlorate has been detected in drinking water from 5.4% of US utilities in at least 26 states and 2 territories. Detectable levels varied from the detection limit of 4–420 μg/l. Perchlorate contamination of the lower Colorado River led to tap water perchlorate levels as high as 24 μg/l (Li et al., 2001). However, more recent work suggests that perchlorate levels in the lower Colorado River may now be lower (3–4 μg/l) due to ongoing mitigation efforts (Sanchez et al., 2006). Using perchlorate-contaminated water for reconstitution of PIF will elevate the overall exposure dose to the infant. In Table 4, we calculated the expected water perchlorate concentration required to reach the perchlorate RfD (0.7 μg/kg per day) for each scenario. Reconstitution of PIF with water contaminated with at least 4 μg/l of perchlorate resulted in the perchlorate RfD's being reached in almost half the dosing scenarios. The significance of this value relates to the EPA's Unregulated Contaminant Monitoring Rule (United States Environmental Protection Agency Unregulated Contaminant Monitoring Program, 2006). This rule requires utilities to report perchlorate levels found in drinking water at concentrations greater than 4 parts per billion or 4 μg/l. The potential health risks associated with exceeding the perchlorate RfD in infants is currently unknown.
Maternal hypothyroidism during pregnancy can adversely affect the cognitive development of the child in the pre- and post-natal period (Haddow et al., 1999; Klein et al., 2001; Pop et al., 2003). Unrecognized congenital hypothyroidism can also lead to mental retardation (American Academy of Pediatrics, 2006). Several ecological and cohort studies have looked at a possible association between perchlorate-contaminated drinking water consumption and TH production dysfunction in infants and children (Lamm and Doemland, 1999; Crump et al., 2000; Feng et al., 2000; Li et al., 2000; Tellez et al., 2005). Of particular interest are two studies that looked at three populations in Chile exposed to different perchlorate concentrations in their drinking water: (1) 100–120 μg/l, (2) 5–7 μg/l, and (3) <4 μg/l (Crump et al., 2000; Tellez et al., 2005). There was no detectable difference in neonatal thyroid function tests or markers of fetal growth retardation (neonatal body weight, length, head circumference) (Crump et al., 2000; Tellez et al., 2005). The findings of these studies are somewhat reassuring, although higher iodine intake in the population may have reduced perchlorate-mediated inhibition of iodide transport. Mean perchlorate concentrations in human breast milk in a Chilean population ranged from 18.3 to 95.6 μg/l. Perchlorate breast milk concentrations in two studies of women in the United States ranged from approximately 1.4–21.4 μg/l (Kirk et al., 2005) and 1.3–411 μg/l (Pearce et al., 2007). It would appear that breast milk-fed infants would also retain the potential for exceeding the RfD (0.7 μg/kg per day) if the same exposure dose estimate calculations were used; however, the lack of any overt, large-scale effect, as evidenced by the aforementioned studies, again is reassuring.
It is also interesting to note that according to the United States Department of Agriculture's Economic Research Service, in 2000 (the latest year for which data were available), Brands D and E in Table 3 commanded 87% of the PIF market share (United States Department of Agriculture—Economic Research Service, 2007). The widespread penetrance of these products, and the potential for utilization of water for reconstitution that has even minimal concentrations of perchlorate, suggest that a significant number of infants consuming bovine milk-based PIFs with lactose will have perchlorate doses in excess of the RfD. Although the perchlorate RfD may be exceeded in some specific situations, the clinical relevance of this is unclear.
Adequate intake of dietary iodide is critical for neurodevelopment in utero and during early life for normal TH production and subsequent growth (National Academy of Sciences, 2004; Pearce et al., 2007). Earlier, albeit limited, research has suggested that in some situations, breast milk may provide inadequate iodine to meet infants' requirements (Kirk et al., 2005; Pearce et al., 2007). The recommended dietary allowance for iodine ranges from 110 to 150 μg in children and can be as high as 290 μg in pregnant women (Kirk et al., 2005). The FDA requires that infant formulas have iodine as an added supplement (among others) at levels between minimum (5 μg) and maximum (75 μg) values per 100 kcal of energy (31). Although the minimum levels of iodine would be insufficient based on exposure modeling (Table 1), it is more likely that the true levels would approach somewhere in between (middle value). In this case, and in situations with higher iodine intakes, no infants would be expected to be iodine deficient. This would likely lessen the possibility of perchlorate-induced thyroid dysfunction, however, the complex relationships among iodine, perchlorate, and their effects on TH production are still being elucidated.
The inability to locate five brands of lactose-free and elemental PIFs called for by the protocol may have affected the results for these types. The results of this study may not be relevant throughout the United States, as all samples were taken within a single city. In addition, our estimated exposure quantities are based on modeling and may not be valid in a real-life setting.
Perchlorate was found in all brands and types of infant formula tested. All bovine milk-based PIFs with lactose have significantly higher concentrations of perchlorate than the other three types tested (soy-based, lactose-free, and elemental). Infants consuming certain bovine milk-based PIFs with lactose may be at risk for exceeding the RfD (0.7 μg/kg per day). Reconstitution of PIFs with water containing perchlorate levels of at least 4 μg/l resulted in 26/48 (54%) hypothetical dosing scenarios exceeding the RfD. Ingestion of PIFs that use water with even minimal amounts of perchlorate for reconstitution could result in a perchlorate dose that exceeds the RfD in many cases. The clinical relevance of exceeding the perchlorate RfD in both an iodide-sufficient and iodide-deficient state are unclear. Further work is needed to clarify both the complex interrelationships among the thyroid, perchlorate, and iodide as well as any potential public health risk with exceeding the perchlorate RfD.