Press Center
Methylmercury Workshop Report - APPENDIX II-A
Response to Questions by the Study Team for the Iraq and the Seychelles Studies
- Chair: Dr. Thomas Clarkson, University of Rochester
- Dr. Conrad Shamleyee, Ministry of Health, Seychelles
- Dr. Donna Palumbo, University of Rochester
- Dr. Christopher Cox, University of Rochester
- Dr. Phillip Davidson, University of Rochester
- Dr. Gary Meyers, University of Rochester
- Dr. Elsa Cernichiari, University of Rochester
SUMMARY OF IRAQ AND SEYCHELLES STUDIES
The Iraq Outbreak
Studies on the Iraq outbreak are still used by regulatory agencies to assess human health risks and may continue to be used in the future. This outbreak, one of the largest mass chemical poisonings ever reported, took place as a result of making bread from seed wheat that had been treated with a methyl mercury fungicide. In the hospitals in Iraq, 600 deaths were recorded and 6000 severe cases of poisoning.
Adult Poisoning
The consumption of the contaminated bread resulted in an insidious and irreversible development of neurological signs and symptoms. No ill effects or even mild discomfort were experienced by the victim during the intake period which lasted for several weeks. After consumption had stopped, the first neurological signs did not appear for over a month. This long latent period is an insidious property of mercury. There were examples in Iraq of individuals who had unknowingly consumed what would prove to be a fatal dose.
Usually the first symptom to appear is paresthesia, a tingling or unusual sensation in the hands and feet and around the mouth. Subsequently more serious signs appear such as ataxia, dysarthria and loss of vision. All these signs and symptoms result from the destruction of neuronal cells in certain areas of the brain. Why such a reactive and mobile form of mercury selectively damages the brain is still a mystery.
Some functional recovery may occur as the victim learns to live with the disability, but the underlying damage is irreversible. It was for this reason the peak mercury level was used as the appropriate measure of dose. For an irreversible poison, the maximum level in the brain should cause the damage, therefore, the peak values were estimated in an exposed group. Often the peak value was estimated from information on the number of loaves consumed, and on measured mercury levels in blood or hair samples. In nearly all cases it was necessary to extrapolate backwards using an assumed half-time for methyl mercury in the body.
In Figure 1 the peak value is plotted on the x axis expressed in units of concentration in hair. Each sign or symptom has its own relationship with the mercury levels. Paresthesia, plotted on the left, has a background frequency in the population of about 5%. At a certain threshold level the frequency rises above background to reach levels affecting the entire population at the highest mercury levels. The more serious signs have higher threshold levels, the fatal one being the highest one of all.
The threshold for paresthesia is the lowest effect level for adult poisoning. This 100 ppm level will be contrasted with the lowest effect level for prenatal exposures.
Prenatal Poisoning
The Iraqi outbreak also gave quantitative risk information on prenatal exposure to methyl mercury. The outbreak in Japan had suggested that the prenatal period was the most sensitive in the life cycle to methyl mercury [Tsubaki and Irukayama, 1977]. Infants with severe brain damage were born from mothers hardly affected by methyl mercury. Subsequent animal experiments also pointed to the sensitivity of the developing brain.
Hair analysis allows us to recapitulate exposures during pregnancy. Hair grows at about 1 cm/ month so that measuring mercury in the hair centimeter by centimeter from the scalp end recapitulates exposure month by month. Thus a sample of maternal hair collected just after delivery provides a history of exposure during pregnancy. In fact, in a recent study in the Seychelles [Cernichiari et al., 1995], it was demonstrated that levels in maternal hair at delivery indicate brain levels in the newborn.
Maternal hair samples collected after the birth of the offspring were used to recapitulate prenatal exposures in the Iraq outbreak. Hair samples were collected from mothers who had been pregnant during the outbreak. Their offspring were examined at about 30 months of age [Marsh et al., 1987]. Examples of very serious brain damage were seen in children similar to what had been seen in Minamata. However, attempts were made to look for less serious and more subtle effects in children that might be expected to occur at lower hair levels. The examinations were conducted under trying circumstances in the deserts of rural Iraq. One of the major challenges was determining the age of a child as there were no birth certificates. Lengthy~~conversations -were made,-through an interpreter, with all members of the extended family in order to learn the age of the child.
The outcome of this prenatal study was a dose response relationship (figure 2) based on the peak level in the mother during pregnancy versus the frequency of delayed walking in the offspring. Delayed walking was defined as not walking until after 18 months of age.
As with the adult picture, one sees a low background frequency unrelated to mercury levels. At higher mercury levels, the frequency rises as the mercury levels in the mother rise. Given the scatter of the points, one cannot identify a precise threshold level but it could be as low as 10 ppm in maternal hair. As indicated earlier, the adult threshold was 100 ppm thus confirming the greater sensitivity of the fetal brain.
These prenatal findings thus raised an urgent question. Ten ppm in hair can be exceeded in fish consuming populations even where mercury has been naturally bioaccumulated in fresh water and ocean fish. Indeed there are examples in Canada where consumption of freshwater fish have produced levels even exceeding 100 ppm in a few cases [Wheatley, 1979]. Given the unusual situation in Iraq of acute exposure as well as other differences, it was imperative to examine fish eating populations for evidence of effects of mercury.
The Sevchelles Studv
Prenatal exposures are of greater concern in fish eating populations. If the estimated threshold from Iraq were indeed as low as 10 ppm in hair, then heavy fish eaters could exceed this value. A prenatal study was undertaken, therefore, in fish eating populations in the Seychelles Islands [Marsh et al., 1995]. As indicated by the data, levels in these populations can exceed 10 ppm.
The Seychelles Islands are situated in the Indian Ocean just south of the equator, about 1000 miles east of Kenya. Over 90% of the population live on the main Island of Mahe. Fish is a highly popular and important dietary item for the people of the Seychelles. It is cheap. Fish consumption averages twelve meals per week.
The excellent medical infrastructure in the Seychelles provided ante-natal care to the expectant mothers as well as important medical history for the study. The study involved, as in Iraq, the measurement of maternal hair levels during pregnancy but much more sophisticated measures of child development. These include a battery of neuropsychological tests appropriate to-the age of the child. In addition, covariates that could affect child development had to be considered
The study took place in two major phases (figure 3). First a pilot group of over 700 infant mothers pairs was established and observed over a five year period. The experience and data gathered from the pilot study was used to plan the more rigorous main study.
In the main study, the children, also a cohort of over 700, were examined at the same age plus or minus two weeks, a very important consideration in any developmental study. The children were examined at a predetermined age as indicated by the vertical arrows. Findings on the pilot group and on the main cohort up to 66 months of age have been published [Davidson et al, 1998]. Anew round of tests are being completed in the children now 9 years of age.
The continuing longitudinal study of the main cohort with appropriate performance tests at age 6, 19, 29 and 66 months provides a consistent picture of healthy children developing according to the standard norms. Specifically, at 66 months of age the following neurodevelopmental tests were applied: the McCarthy Scales of Children's Abilities, the Preschool Language Scale, the Woodcock-Johnson Applied Problem and Letter and Word Recognition Achievement Tests, the Bender Gestalt Test, and the Child Behavior Check List. No adverse developmental effects of either pre- or postnatal exposure to methyl mercury could be detected.
The main study is continuing. By the Spring of 1999, the main cohort testing at 96 months will be completed. The 96 months battery includes neuropsychological and developmental tests. Anew battery including sophisticated neuropsychological endpoints, is being field tested now on the pilot cohort.
References
Bakir, F., S .F. Damluji L. Amin-Zaki, M. Murtadha, A. Khalidi, N.Y. Al-Rawi, S. T1kriti, H.I. Dhahir, T. W. Clarkson, J.C. Smith and R.A. Doherty, Methylmercury poisoning in Iraq. Science, 181 (1973) 230-241.
Cernichiari, E., R. Brewer, G.J. Myers, D.O. Marsh, L.W. Lapham, C. Cox, C.F. Shamlaye, M. Berlin, P.W. Davidson andT.W. Clarkson, Monitoring methylmercury during pregnancy: maternal hair-predicts fetal brain exposure. Neurotoxicology, 16 (DISC) 705-710.
Cox, C., T.W. Clarkson, D.O. Marsh, L. Amin-Zaki, S. T1kriti, and G.J. Myers., Dose-response analysis of infants prenatally exposed to methyl mercury: An application of a single compartment model to single-strand hair analysis. Env. Res. 49 (1989) 318-332.
Davidson, P.W., G.J. Myers, C. Cox, C. Axtell, C. Shamlaye, J. Sloane-Reeves, E. Cernichiari, L. Needham, A. Choi, Y Wang. M. Berlin and T.W. Clarkson, Effects of prenatal and postnatal methylmercury exposure from fish consumption on neurodevelopment. Outcomes at 66 months of age in the Seychelles Child Development Study. J. Am. Med. Assoc., 280 (1998) 701-707
Marsh, D.O., T.W. Clarkson, C. Cox, G.J. Myers, L. Amin-Zaki and S. A1 Tlkriti, Fetal methylmercury poisoning: relationship between concentration in single strands of maternal hair and child effects. Arch. Neurol., 44 (1987) 1017-1022.
Marsh, D.O. , T. W. Clarkson, G.J. Myers, P.W. Davidson, C. Cox, E. Cernichiari, M.A. Tanner, W., C. Shamlaye, O. Choisy, C. Hoareau and M. Berlin, The Seychelles study of fetal methylmercury exposure and child development: Introduction. Neurotoxicology, 16 (1995) 583-596.
Tsubaki, T and K. Irukayama, Minamata Disease, Elsevier, Amsterdam, 1977, pp. 1 -317.
Wheatley, B. Methylmercury in Canada, Ministry of National Health and Welfare, Medical Services Branch, Ottawa, Canada, 1979.
Figure Legends
Figure 1. The frequency of neurological signs and symptoms is plotted against the maximum (peak) hair concentration in a population of adults exposed to a methyl mercury fungicide in home-made bread [Bakir et al., 1973]. This 'hockey-stick' graphical representation assumes there is a background frequency unrelated to mercury exposure indicated by horizontal lines. This is followed by an increasing frequency as the effects of methyl mercury become detectable. The point of intersection of the horizontal and inclined lines is taken as a "practical" threshold limit for each sign and symptom. Thus f1gure 1 indicates that the threshold for the earliest symptom, paresthesia, is approximately 100 ppm in hair.
Figure 2. The frequency (% response) of motor retardation in prenatally exposed infants is plotted against the maximum maternal mercury level during pregnancy. The total cohort was 83 infant-mother pairs exposed during the Iraq outbreak. Motor retardation was defined as failure to walk at 18 months of age (Marsh et al., 1987).
The 'hockey stick' graphical analysis indicated as background frequency of 0%. This is probably an artifact of the low number of data points. A high background frequency would raise the estimated practical threshold, here indicated as 10 ppm (from Cox et al., 1989)
Figure 3. The overall plan of the Seychelles Child Development Study. Apilot cohort of over 750 infant-mother pairs was established over a two year period starting in 1985. The main cohort of roughly the same size was established over a one year period starting in 1989. The arrows indicate the dates when neurodevelopmental tests were made. The pilot cohort was used from a cross-sections study whereas the main cohort was designed as a double-blind prospective longitudinal study (Marsh et al., 1995).
Seychelles Child Development Study responses to questions:
- What are the exposures to organic or inorganic mercury?
- What are the sources of exposure?
- Details regarding Hg levels
- What is the specificity and sensitivity of health endpoints.
- What are the most specific and sensitive tests.
- How are tests impacted by cultural and behavioral practices
- Are the developmental tests comparable across different studies?
- What are the confounders that affect health endpoints positively (e.g. selenium omega 3 fatty acids and negatively (e.g. PCB, alcohol use) health conditions.
- What is the variability within and across populations and host factors.
- For each epidemiological study, which features-are particularly stmng or weak?
No local pollution is known. Population is exposed to methyl mercury mainly from fish. Marine mammals are not consumed.
The Seychelles has an active dental program. As in other countries, mothers do have dental amalgam fillings.
Hair mercury values - Main Study Dataset for analysis
| # of SAMPLES | MERCURY LEVELS ng/mg hair | ||||
|---|---|---|---|---|---|
Mean |
Std Dev. |
Lowest |
Highest |
||
Maternal |
711 |
6.8 |
4.5 |
0.5 |
27 |
Children |
|||||
at 6 months |
699 |
6.6 |
4.4 |
0.5 |
35 |
at 19 months |
739 |
4.8 |
3.1 |
0.3 |
23 |
at 66 months |
711 |
- 6.5 |
3.3 |
0.9 |
26 |
What were the peak hair Hg levels during pregnancy?
The definition of peak levels depends on how you measure mercury levels in hair. For example in Iraq, the mercury levels were sufficiently high to measure single strands in 2 mm segments by an X-ray fluorescence (XRF) technique. The nature of the exposure was such (an intake period followed by cessation of further exposure) that a pharmacokinetic model could be applied to the data to calculate the true peak level. Thus the estimated peak value was based on many data points (for details, see Cox et al. 1989).
In other cases such as the Seychelles, Peru, New Zealand and Canada, 1 cm segments of hair were analyzed over a 10 cm length corresponding to the period of pregnancy or longer segments if seasonal changes were the subject of interest (e.g. Canada, Phelps et al. 1980). The highest one centimeter segment was taken as the 'peak' value.
Lengths of hair corresponding to each trimester of pregnancy were measured in the Seychelles study (see Cernichiari et al 1995) but they have not been used to estimate peak values. Mercury levels in the trimester were highly correlated.
Kjellstrom et al (1989) reported that the peak 1 cm segment was between 1.4 and 1.6 times the average mercury levels for the whole period of pregnancy. Our finding on the Seychelles sample support this conclusion ¿
In general, in ocean fish consuming populations in New Zealand, The Seychelles and Peru, the highest and lowest 1 cm levels during pregnancy did not differ by more than a factor of two (Clarkson, 1997).
Is cord tissue a reliable indicator of PCB exposure?
Were cord and blood PCB measured?
What were PCB analytical techniques and congeners analyzed and the tissue used?
Twenty-eight PCBs congeners were measured in each sample:
PCB 101, 105, 110, 118, 138, 146, 152, 156, 170,172, 178, 180, 183, 187, 189, 193, 194, 195, 201, 203, 206, 28, 52, 56, 66, 74, and 99.
What were urinary levels of inorganic mercury in suckling infants.
Were data separated temporally as prepartum or by trimester or month for consumption of Hg from food or by Hg analysis of samples?
Were measurements of methylmercury obtained?
Provide details on maternal hair levels if not published.
Provide details of hair levels during pregnancy and statistical associations with endpoints if not published.
Specificity refers to a test's capacity to assess the function it was designed for, minimizing interfering influences by other functions. It is the corollary of construct validity. Sensitivity refers to the test's capacity to detect change in the function in question. Test validity and sensitivity can be ascertained by standardization. All properly standardized tests possess latent sensitivity viz the reference population. Manifest sensitivity will be related to sample size, power, and the clinical relevance of a difference in scores presumably caused by some external factor, such as an exposure to a neurotoxicant. Sensitivity for developmental and neuropsychological tests has not been established for methyl mercury or other neurotoxicants. For example, detection of effects does not in itself demonstrate sensitivity. klidity must be established based upon demonstrating a high correlation between scores obtained by the reference population and their scores on other measures known to access the domain or function in question.
The SCDS defined developmental domains to be assessed based upon previous literature (Davidson et al., 1995). Tests were then chosen that were known to possess validity to measure each of these domains at different ages. We used standardized developmental tests at each age. The sample size for the main study was ample enough to detect a five point difference on any test with a standard mean of 100 (SD=15). This degree of sensitivity was selected based on MeHg effect sizes reported earlier. This is the same standard as was adopted by researchers when the effects of lead exposure were studied.
As children develop, cognition, language, perception, and social behavior all become more complex and differentiated. Thus the types of tests that might be considered the most specific for a particular function will vary with the age of the child. For example, infants are capable of many functional behaviors, but there is a limited spectrum of acceptable tests available to assess those behaviors. Likewise, many neuropsychological functions may not be well enough developed in toddlers at the stage of concrete operations to be tested by traditional child neuropsychological tests.
All psychological tests are affected by cultural and behavioral practices, some more than others. Thus, when using a test in a culture or setting that differs from the standardization sample for that test, performance characteristics for the test must be established prior to interpretation. In the SCDS, each test was piloted extensively in the Seychelles, and its performance characteristics determined to be comparable to what would be expected for U.S. children before it was included in the main study battery (Davidson et al., 1995). In fact, some tests that had been tentatively selected for use were excluded from the main study after pilot testing revealed less than nominal performance characteristics.
In general, there are certain conditions under which tests are comparable across studies. The tests must be valid, reliable, given under standardized conditions at appropriate ages and must assess similar neurodevelopmental domains.
We are assuming a "confounder" is a factor that varies both with the dependent and independent variable, e.g., PCB and omega-3 fatty acids and length of breast feeding. Covariates are those factors that may affect performance but do not necessarily co-vary with the independent variable e.g. alcohol consumption and diabetes.
What is the general health status.
Details on co-variates/confounders.
We have measured omega 3 fatty acids and selenium in a randomly chosen group of cord blood samples. Both are in the normal range, according to the literature. These data have not been published.
We assume this question applies to studies which we have conducted. The greatest differences would be between the Iraq (Bakir et al., 1973) and the Seychelles populations (Shamlaye et al., 1995). The Peruvian population (Marsh et al., 1995b) is similar to the Seychelles.
We assume this question pertains to studies that we have conducted and that the "methyl mercury group" will evaluate all of the studies. We will respond to the Seychelles study as the Iraq study has been reviewed extensively (e.g. WHO, 1990).
What were the exclusion criteria.
Were outlying data points considered.
Were undocumented environmental and cultural factors not accounted for.
With the benefit of hindsight, identify covariates that should have been included.
Beneficial nutritional constituents of fish may play a role in determining the outcome of these studies. For example, the omega-3 fatty acids, especially DHA, were first shown to have beneficial effects on brain development in pre-term infants. More recent studies have shown that these beneficial effects apply to full term infants. Plasma levels of omega-3 fatty acids are known to parallel fish intake. This in turn determines mercury levels in hair.
High fish intake also lead to higher selenium levels in blood. It has been argued that the high intake of protein in fish eaters may reduce brain uptake of methyl mercury.
Even beyond the nutritional role of fish, other dietary factors must be considered. Toxicologists are now aware that the outcome of toxicity tests is substantially affected by caloric intake.
It is now clear that studies such as our study in the Seychelles should be accompanied by detailed dietary information not only with regard to fish but overall dietary patterns.
The Seychelles Child Development Study was designed as a confirmatory study. This means that the goal of the study was to test specific hypotheses concerning the statistical associations between fetal exposure to methyl mercury, as measured by the maternal hair level, and developmental outcomes. That is, the model for the study was a clinical trial, rather than an exploratory, hypothesis generating study. Because this was an observational study, it was necessary to control for confounding variables in the statistical analysis. At the same time, a considerable amount of background information was available concerning variables associated with performance on developmental tests that could constitute potential confounders. This information was used to develop a full list of confounding variables that could be assessed in Seychelles. To reduce the possibility of overadjustment, confounders known to be most likely to affect child development were retained in a reduced list. These two lists of confounders were used to develop full and reduced regression models, which were used, with some modification as the children became older, for all of the endpoints in the study. Since earlier studies, including Iraq, had found differential effects in males and females, each regression model was run with and without separate mercury effects for males and females (a mercury x gender interaction). This resulted in a total of four regression models for each endpoint. In addition statistical outliers were removed from each analysis and the model was rerun. The definition of outlier was a standard one made a priori. Adescription of this analysis plan has been published. Although we were careful to limit the number of correlations computed for the analysis of each individual endpoint, no adjustment has been made for the number of endpoints analyzed. This number has become rather large, and from a purely statistical point of view the interpretation of our results should include some adjustment for the number of analyses which we have performed.
In addition to prenatal exposure, we have also begun to examine the effects of postnatal exposure. Analyses for the 66 month tests included variables measuring both prenatal and postnatal exposure. Prenatal exposure was assessed using the maternal hair level and postnatal exposure was measured by the level in a sample of hair taken from the child at the time of testing. The unexpected finding of associations between postnatal exposure and enhanced development has generated an a posteriori hypothesis concerning the possible beneficial effects of fish consumption. In addition, measurements have now been completed for samples of hair taken from the children at 6 and 19 months of age. While we do have both toxicological and nutritional hypotheses concerning the effects of postnatal exposure, we have no specific hypotheses as to which of these three child hair samples (6, 19 and 66 months) should be most strongly associated with developmental outcomes, and some additional discussion will be needed before final statistical analyses can begin.
REFERENCES
Bakir F. Damluji SF, Amin-Zaki L, Murtadha M, Khalidi A, Al-Rawi NY, Tikriti S. Dhahir HI, Clarkson TW, Smith JC and Doherty RA. Methylmercury poisoning in Iraq. Science 181:230-141, 1973.
Cernichiari E, Toribara TY. Liang L, Marsh DO, Berlin M, Myers GJ, Cox C, Choisy O. Davidson PW and Clarkson TW. The biological monitoring of mercury in the Seychelles study. Neunvtoxicology 14(4) :613-628, 1995.
Clarkson TW. Methylmercury: Loaves versus fishes. CUTActivities 17(5):2-7, 1997.
Cox C, Clarkson TW, Marsh DO, Amin-Zaki L, A1 Tikriti S. Myers G. Dose-response analysis of infants prenatally exposed to methylmercury: an application of a single compartment model to single-strand hair analysis. Environ. Res. 49:318-322, 1989.
Davidson PW, Myers GJ, Cox C, Axtell C, Shamlaye C, Sloane-Reeves J. Cernichiari E, Needham L, Choi A, Wang X Berlin M and Clarkson TW. Effects of prenatal and postnatal methylmercury exposure from fish consumption at 66 months of age. JAMA 280(8):701707, 1998.
Davidson PW, Myers GJ, Cox C, Shamlaye C, Tanner MA, Choisy O. Sloane-Reeves J. Cernichiari E, Marsh DO and Clarkson TW. Neurodevelopmental test selection, adrninistration and performance in the main Seychelles Child Development Study. Neurotoxicology 16(4):665-676, 1995.
Kjellstrom T. Kennedy P. Wallis S. Stewart A, Friberg L, Lind B. Wutherspoon T. Mantell C. Physical and mental development of chidren with prenatal exposure to mercury from fish. Stage 2: interviews and psychological tests at age 6. The National Swedish Environmental Pnvtection Board, S- 171 85 Solna, Sweden, 1989, p. 38, Table 6.
Marsh DO, Clarkson TW, Myers GJ, Davidson PW, Cox C, Cernichiari E, Tanner MA, LednarW, Shamlaye C, Choisy O. Hoareau C, and Berlin M. The Seychelles study of fetal methylmercury exposure and child development: Introduction. Neurotoxicology 16(4):583-596, l995a.
Marsh DO, Turner MD, Smith JC, Allen P. and Richdale N. Fetal methylmercury study in a Peruvian fish-eating population. Neunvtoxicology 16(4):717-726, l995b.
Phelps RW, Clarkson TW, Kershaw TG, Wheatley B. Interrelationships of blood and hair mercury concentrations in a North American-population exposed to methylmercury. Arch. Environ. Health, 35(3): 161-168, 1980.
Shamlaye C, Marsh DO, Myers GJ, Cox C, Davidson PW, Choisy O. Cernichiari E, Choi A, and Clarkson TW. The Seychelles Child Development Study on neurodevelopmental outcomes in children following in utero exposure to methylmercury from a maternal fish diet: Background and demographics. Neurotoxicology 16(4):597-616, 1995.
World Health Organization. Environmental Health Criteria 101. Methylmercury. WHO, Geneva, 1990
Return to Workshop Table of Contents
