To the Editor:
In their article in the December 2010 issue of JOEM, Steinmaus, Miller, and Smith,1 reevaluate the research of Buffler et al2 and find, contrary to the original findings, that there was a positive association between the thyroid function of newborns during the period of 1997 to 1998 and whether the mother lived in communities that were potentially exposed to perchlorate in drinking water with at least 5 μg/L or not. Perchlorate is of course a goitrogen, an agent that prevents the thyroid from properly absorbing and processing iodide. If an individual is exposed to a sufficient quantity of one or more goitrogens over a long enough time, iodide starvation will result which can produce goiter (hyperplasia of the thyroid), hypothyroidism (elevated TSH concentrations), and hypothyroxinemia (depressed concentrations of T4). Maternal hypothyroxinemia is known to produce neurodevelopmental deficits among newborns. So, the import of this research is clear.
Nevertheless, there are a number of problems that would suggest that the conclusions of this study are incorrect. First, there is the issue of relative source contribution of perchlorate. As the authors rightly note, perchlorate exposure in the United States is entirely ubiquitous based on a number of studies, basically every person tested shows perchlorate exposure even in areas with no known perchlorate in the drinking water.3–5 It is now well established that the majority of perchlorate exposure is from food and not water.6 Even if the original classification of communities into more likely and less likely to be exposed to perchlorate via drinking water was accurate, it is highly improbable that there was any difference between these communities in the quantities of perchlorate ingested via food. With dietary sources contributing a large majority to the overall exposure, it is fairly safe to say that the contribution from drinking water was small at best.
As the authors correctly note, there are a number of other goitrogens in the environment, which could also cause the reported elevated TSH, most notably nitrate, cyanide, and thiocyanate. The authors write: “Importantly though, to cause significant confounding, these factors would have to be associated with both thyroid hormones and perchlorate exposure, and these associations would have to be fairly strong.” As it turns out, this is exactly the case for nitrate. In 2007, an article was published demonstrating that in California drinking water sources contained thousands of times more nitrate than perchlorate on average and that nitrate was much more ubiquitously distributed.7 During period of 1994 to 2003 in California, 31,464 samples collected for perchlorate analysis but only 6125 were reported above the reporting limit of 4 μg/L whereas 204,386 samples collected for nitrate analysis of which, 174,373 had values greater than the reporting limit of 2 mg/L (2000 μg/L).
The median concentration of nitrate in California in this study period was 16,000 μg/L (the average was 21,000 μg/L so it data is clearly right skewed and nonnormal). During the same period, the median perchlorate concentration value was less than 4 μg/L, that is, not detected (the average was 3.6 μg/L so this data was also right skewed and nonnormal). Nevertheless, many water sources were sampled for both analytes during the period of 1997 to 2003 within 180 days of each other and among those sources with a perchlorate result above the reporting limit, the median perchlorate concentration was 7.3 μg/L and the median concentration of nitrate was 31,000 μg/L. What this means is that, on average, water samples with perchlorate above 4 mg/L had more than twice as much nitrate as those with perchlorate less than 4 μg/L. In fact, among water samples with perchlorate, there was a weak but statistically significant correlation between the concentration perchlorate and nitrate (Mann-Whitney rank sum test R2 = 0.31, P < 0.001). So, if the authors had correctly classified communities into those has having an average concentration of perchlorate in their drinking of either less than or more than 5 μg/L they have also mostly likely divided them into communities having either 16,000 or 31,000 μg/L nitrate in their drinking water. So, if the authors are correct in regards to hypothesized association between the community of maternal residence and infant TSH concentration, nitrate in drinking water is a far more likely candidate than perchlorate. Of course, nitrate is a significantly less potent goitrogen than perchlorate, Tonacchera et al8 reports perchlorate as 240 times more potent than nitrate. Nevertheless, even if it were very conservatively assumed that nitrate were 1000 times less potent than perchlorate, individuals in this study drinking water with more than 5 μg/L are likely consuming three to six times that much nitrate on a “perchlorate equivalent concentration” (PEC) basis.
In support of their conclusions, the authors cite two articles derived from the 2001 to 2002 NHANES study which showed “...evidence of a linear association between increasing urinary perchlorate and decreasing serum T4 were identified in women in NHANES with low iodine and high thiocyanate exposures, and median perchlorate intakes of about 4 μg/day...”. This is certainly true as Blount et al write “For women with urinary iodine less than 100 mg/L, multiple regression analysis found perchlorate to be a significant predictor (P < 0.0001) of T4 with a coefficient for log perchlorate of −0.8917.” Nevertheless, Bount et al also note “For women with urinary iodine at least 100 μg/L, perchlorate was not a significant predictor of T4 (P = 0.5503) but remained a significant predictor of log TSH (P = 0.0249)” that suggests that the women in this study fell into two subpopulations, a smaller, sensitive one with lower iodide intake (N = 348) and a larger, more robust subpopulation with higher iodide intake (N = 724). What is interesting is that while urinary nitrate concentrations were not a predictor for T4 in the smaller sensitive subpopulation, it was a significant predictor of T4 (coefficient −1.1215, P = 0.0249) for the larger robust subpopulation with adequate iodide intake. If this data is taken a face value, nitrate is a far greater threat to the thyroid health of women and newborns. Certainly, the concentrations of nitrate were many times larger in this study (geometric mean = 38,000 μg/L) as compared to perchlorate (geometric mean = 2.8 μg/L), even on PEC basis. However, the prima facie reading of these results could be misleading as there were only 21 women who were actually hypothyroxinemic, the critical health end-point. Furthermore, the NHANES results do not indicate that those hypothyroxinemic women were iodide deficient, that is, their condition was caused by iodide starvation brought on by excess goitrogen exposure. In other words, the NHANES data would suggest that if there is some association between exposure to goitrogens, such as perchlorate and nitrate, and thyroid hormone concentrations, the changes are entirely within the normal clinical ranges, that is, they were not sufficient to produce pathological changes.
Quite aside from this, there is also the question relative source contribution of nitrate and other goitrogens. It is well known that 90% of nitrate consumed comes from dietary sources, not drinking water.9 Furthermore, there are numerous other dietary goitrogens not found in drinking water such as thiocyanate, which is more than 10 times more potent than nitrate as a goitrogen.8 It is a metabolite of cyanogenic glucosinolates that occurs naturally in many vegetable products such as cabbage (and related species of Brassica), cassava, sweet potatoes, corn, apricots, cherries, and almonds.10 Blount reports that the geometric mean of the concentration of thiocyanate in the NHANES study was 1200 μg/L, hundreds of times higher than the geometric mean perchlorate concentration. There are other organic goitrogens such as isoflavones found in soy and goitrin found in turnips. Thus, if the entire body burden of all goitrogens is considered, the amount contributed by perchlorate in drinking during the period of 1997 to 1998 would have to be trivially small in comparison.
In conclusion, the hypothesis of this article, that differences in TSH concentrations found in infants born in 1998 to 1999 based on maternal residence in communities thought to be at risk for perchlorate exposure through drinking water cannot be supported because of the strong cooccurrence of nitrate and perchlorate in California groundwaters and because of the relative contribution of nitrate and perchlorate from food as opposed to tap water. Furthermore, given huge load of other goitrogens people in the study areas would have been consuming from food during the study period, any variation caused by differences in tap water perchlorate, or even nitrate consumption, would be infinitesimally small.
David Eugene Kimbrough, PhD
Water Quality Manager, Pasadena, CA
1. Steinmaus C, Miller MD, Smith AH. Perchlorate in drinking water during pregnancy and neonatal thyroid hormone levels in California. 2010;52(12):1217–1224.
2. Buffler PA, Kelsh MA, Lau EC, et al. Thyroid function and perchlorate in drinking water: an evaluation among California newborns, 1998. Environ Health Perspect. 2006;114:798–804.
3. Blount BC, Pirkle JL, Osterloh JD, Valentin-Blasini L, Caldwelldoi KL. Urinary Perchlorate and Thyroid Hormone Levels in Adolescent and Adult Men and Women Living in the United States. Environ Health Perspect. 2006;114:1865–1871.
4. Oldi JF, Kannan K. Analysis of perchlorate in human saliva by liquid chromatography-tandem mass spectrometry. Environ Sci Technol. 2009;43(1):142–147.
5. Pearce EN, Leung AM, Blount BC, et al. Breast milk iodine and perchlorate concentrations in lactating Boston area women [published online ahead of print February 20, 2007]. J Clin Endocrinol Metabolism. doi:10.1210/jc.2006-2738.
6. Huber DR, Blount BC, Mage DT, Letkiewicz FL, Kumar A, Allen RH. “Estimating perchlorate exposure from food and tap water based on US biomonitoring and occurrence data [published online ahead of print June 23, 2010].” J Exposure Sci Environ Epidemiol. doi:10.1038/jes.2010.31
7. Kimbrough DE, Parekh P. The occurrence & co-occurrence of perchlorate and nitrate in California drinking water sources. J Am Water Works Assoc. 2007;99:126–132.
8. Toncacchera M, Pinchera A, Dimida A, et al. “Relative potencies and additivity of perchlorate, thiocyanate, nitrate, and iodide on the inhibition of radioactive iodide update by the human sodium iodide symporter. Thyroid. 2004;14(12):1012.
9. National Research Council. “Nitrate and Nitrite in Drinking Water—Chapter 4 Exposure Assessment,” Commission on Life Sciences, Committee on Toxicology, Subcommittee on Nitrate and Nitrite in Drinking Water, 1995 ISBN-10: 0-309-08370-2.
10. VanEtten CH, Daxenbichler ME, Wolff IA. “Natural glucosinolates (thioglucosides) in foods and feeds.” J Agric Food Chem. 1969;17(3):483–491. doi: 10.1021/jf60163a013.