Prenatal Mercury Exposure and Neurodevelopment up to the Age of 5 Years: A Systematic Review

Round 1

Reviewer 1 Report

This manuscript is tried to review the neurodevelopmental delays or effects of young children after they were exposed to mercury. Actually, I recommend to do a lot of modification after I read this review article, although I give it the positive impact. Please see my comments as the below.

1. Keywords and abstract: I hope the authors make them clear. Form the abstract, the significant findings and results are required. It is very difficult to let the audience understand the main adverse effects of neurodevelopment after the young children were exposed to mercury.
2. Introduction: The authors did not explain the environmental fate of mercury and how the mercury is biaccumulated at young children via the environmental matrices from the environment. 
3. Introduction: the analytical instruments for detection of mercury in human specimens and the methods for the young children's neurodevelopment are required.
4. Methods: the  authors shall need to define their age form their selected published articles.
5. Results: The different age children use the different scales or methods. The authors shall explain or define the scales in their selected articles. Or Making tables to explain their differences.
6. What is BSID-III? Do the authors mean the Bayley-III scales? Do the authors know what is different between BSID-II and Bayley-III? What is the different between K-BSID-II and BSID-II?
7. WPPSI-R??? Do the authors explain their version of WPPSI-R?
8. The authors use the several scales from several papers. I want to know what is the significant results of neurodevlopment after the young children were exposure to mercury.
9. What is the different among the different stages of young children like the newborns, infants, toddlers, pre-school children, school children, or adolescents? Do they have different adverse effects in the different stages?

Author Response

Thank you for considering our paper. We have made adjustments which we hope meet your concerns and suggestions.

 

1. Keywords and abstract: I hope the authors make them clear. Form the abstract, the significant findings and results are required. It is very difficult to let the audience understand the main adverse effects of neurodevelopment after the young children were exposed to mercury.

We were not able to meta-analyse the studies to give one overall estimate due to the substantial heterogeneity in the study methods. Instead, we have aimed in this paper to provide a systematic review of the evidence, which considers and summarises all results rather than only those which were statistically significant. Previous reviews of mercury & neurodevelopment (e.g. Bellinger et al, 2018) did not include all relevant studies and tended to focus on significant findings. This may give the reader the false impression that there is strong overall evidence for an adverse effect from mercury. However, considering all results from studies in this review we did not find support for this.

We have slightly modified the abstract to make this clearer [Lines 24-31].

 

 

2. Introduction: The authors did not explain the environmental fate of mercury and how the mercury is bioaccumulated at young children via the environmental matrices from the environment.

The introduction has been rewritten to give a better explanation of the path from environmental emissions to fetal accumulation. [Lines 48-63]

 

3. Introduction: the analytical instruments for detection of mercury in human specimens and the methods for the young children's neurodevelopment are required.

Results: The different age children use the different scales or methods. The authors shall explain or define the scales in their selected articles. Or Making tables to explain their differences.

The instrument details for analysing mercury are available in Supplementary File Part 6, and a summary of each neurodevelopmental scale has been added to Supplementary File Part 5. A detailed comparison  of the differences between scales was beyond the scope of this review, but references are provided which contain further details of each scale.

 

4. Methods: the authors shall need to define their age form their selected published articles.

The abstract and methods have been updated to make the age range (0 to 5 years old) clearer. [Line 20, Table 1]

The age range used in each study is available in the results tables which have been moved from the supplementary file into the main text.  [Tables 4-9]

 

5. What is BSID-III? Do the authors mean the Bayley-III scales? Do the authors know what is different between BSID-II and Bayley-III? What is the different between K-BSID-II and BSID-II?

There are papers in the review which refer to the Bayley Scales of Infant and Toddler Development–Third Edition both as “BSID-III” and “Bayley-III”. Most refer to it as Bayley-III, so the name has now been changed, with a note added that some studies use “BSID-III”. [Lines 172-173]

K-BSID-II is a Korean language version of the BSID-II, and a reference has been added to Supplementary File Part 5.

 

6. WPPSI-R??? Do the authors explain their version of WPPSI-R?

The authors Lederman et al (2008) do not specify a version of the WPPSI-R, only that it is the WPPSI-R (revised) and not the WPPSI.

 

7. The authors use the several scales from several papers. I want to know what is the significant results of neurodevlopment after the young children were exposure to mercury.

This paper aims to summarise all results to assess the overall strength of the evidence. A small proportion of study results identified evidence of an association, but these did not appear to be specific to a particular neurodevelopmental outcome, measurement method, or child age. Our overall finding is therefore that there is mostly no strong evidence of an association between neurodevelopment and children.

The discussion has been modified to make this finding clearer [Lines 464-473]

 

8. What is the different among the different stages of young children like the newborns, infants, toddlers, pre-school children, school children, or adolescents? Do they have different adverse effects in the different stages?

We examined the results for differences between different age groups, but did not find any. This has been added to the abstract [Lines 25-26] and discussion [Lines 380-385]

Reviewer 2 Report

Overall this paper provides critical considerations for development of appropriate statistical considerations for evaluating the role of mercury on development of children and should be published with the following recommendations. 

Very early on (lines 50 to 60) the authors should explain that methyl mercury is by far the most toxic and they give the false impression that all forms of mercury are equally toxic.  They fail to point out that the methylation process is critical to the global exposure issue. They say “The most common and readily absorbed forms of dietary mercury are elemental Hg and organic Hg in the form of methylmercury (MeHg), which may be harmful to fetal neurodevelopment because both are lipid soluble and can cross the placenta and blood-brain barrier [13,14]”. In the natural food web there is very little exposure to elemental mercury. This is more of an industrial exposure and it is not very toxic anyhow. Another exposure to elemental mercury is dental amalgams. They also state that these forms of mercury are lipid soluble. This is a common misconception that exists in some of the older literature but methyl mercury is mostly association with protein and sulphur containing amino acids not lipid. There is little discussion about oxidized mercury (Hg2+) which is the common form in rain and aquatic systems. While it is true that elemental mercury is the form carried in long range transport, it must be oxidized to return to earth and soluble in rain. This section is the first red flag that these authors know little about global cycling of mercury. They also ignore much of the literature on mercury exposure in First Nations communities (eg Mergler et al. plus many others where values are much higher than that considered here).

In lines 46 to 48 it states “For those not at risk of occupational contact with mercury, the environment and diet are the most common sources of exposure”. The environment is of little consequence as almost all methyl mercury exposure is from eating fish and seafood. The exception is exposure due to mercury amalgams and exposure in vaccines.

Lines 58 to 61 the authors say “Compared with adults, the developing infant has both higher levels of mercury crossing the blood-brain barrier and less efficient clearance [7,15,16], which may lead to the accumulation of harmful concentrations in the infant brain [17,18] during a potentially critical window of development [19]”. This is correct but they are missing the contributions of many other references such as Sakamoto and colleagues who showed that by the time a baby is born it is more contaminated with mercury than the mother.

Lines 76 to 78 the authors say “Fish are a source of both mercury and of essential nutrients hypothesised to be involved in reducing the likelihood of autism such as long-chain poly-unsaturated fatty acids (LCPUFA) [22,23]”. Again the authors should make it clear that the main exposure problem is with methyl mercury.

Lines 185-186 It states “The mean Hg in maternal whole blood ranged from 0.64 [39] to 3.71 μg/L, and maternal hair sample concentrations from 0.3 to 5.7 μg/g”. This may be the smoking gun for the discrepancy between this and other literature. The values selected are really quite low. Hair values of 20 to 50 ug/g should have been included and of course there is a safety factor involved when scaling down to the low values considered in this analysis. The WHO Safe limit for children and women of childbearing age is 2 ppm. USEPA safe levels are even less.

Lines 435-436 gives the statement “The evidence for an association between prenatal mercury concentrations and neurodevelopmental functioning below the age of 5 years old is weak”. This reviewer suggests adding “in the range of concentrations considered in this analysis”. The bottom line is they looked at levels near the safe level and so seeing a weak association would be expected.

Two confounding factors which were not addressed in this review was whether or not the mothers had mercury amalgam fillings in their teeth or if the young children were given vaccines containing mercury. These problems have not been approached responsibly as many researchers have had vested interests in the outcomes. In vaccines the mercury is in the form of trimethyl mercury called Thermarasil and the chemical methods used were not reliable. 

 

Author Response

Thank you for considering our paper, and for your positive and constructive feedback. We have made adjustments which we hope meet your concerns and suggestions.

 

1. Very early on (lines 50 to 60) the authors should explain that methyl mercury is by far the most toxic and they give the false impression that all forms of mercury are equally toxic. They fail to point out that the methylation process is critical to the global exposure issue. They say “The most common and readily absorbed forms of dietary mercury are elemental Hg and organic Hg in the form of methylmercury (MeHg), which may be harmful to fetal neurodevelopment because both are lipid soluble and can cross the placenta and blood-brain barrier [13,14]”. In the natural food web there is very little exposure to elemental mercury. This is more of an industrial exposure and it is not very toxic anyhow. Another exposure to elemental mercury is dental amalgams.

They also state that these forms of mercury are lipid soluble. This is a common misconception that exists in some of the older literature but methyl mercury is mostly association with protein and sulphur containing amino acids not lipid. There is little discussion about oxidized mercury (Hg2+) which is the common form in rain and aquatic systems. While it is true that elemental mercury is the form carried in long range transport, it must be oxidized to return to earth and soluble in rain. This section is the first red flag that these authors know little about global cycling of mercury.

We have modified the introduction to focus primarily on methylmercury, given that it is the more common dietary exposure and potentially more toxic. While industrial exposure and Hg found in dental amalgams and vaccines are of interest, our literature search was primarily concerned with exposure during pregnancy, which for the most part we expect to be dietary (although dental care and the chances of treatment may increase).

The role of oceanic methylation has been added.

The biological pathways of absorption have been removed because there are constraints on the size of the introduction, and it would not have been possible to give a satisfactory explanation in the limited space available. This paper has an epidemiological focus so while the biological mechanisms are important, we do not feel they are essential to understand the primary findings of this review.

Mercury as discussed in the introduction has been largely rewritten [Lines 47-63]. However, the length of the introduction is a limiting factor to how much detail it was possible to add.

 

3.They also ignore much of the literature on mercury exposure in First Nations communities (eg Mergler et al. plus many others where values are much higher than that considered here).

We could not see any studies in the First Nations communities which met our inclusion criteria, but the high levels of mercury identified in this population have been noted in the discussion. [Lines 483-484]

 

4. In lines 46 to 48 it states “For those not at risk of occupational contact with mercury, the environment and diet are the most common sources of exposure”. The environment is of little consequence as almost all methyl mercury exposure is from eating fish and seafood. The exception is exposure due to mercury amalgams and exposure in vaccines.

These lines have been removed because they appeared redundant following changes to the rest of the introduction.

 

5. Lines 58 to 61 the authors say “Compared with adults, the developing infant has both higher levels of mercury crossing the blood-brain barrier and less efficient clearance [7,15,16], which may lead to the accumulation of harmful concentrations in the infant brain [17,18] during a potentially critical window of development [19]”. This is correct but they are missing the contributions of many other references such as Sakamoto and colleagues who showed that by the time a baby is born it is more contaminated with mercury than the mother.

This detail has been added. [Lines 65-66]

6. Lines 76 to 78 the authors say “Fish are a source of both mercury and of essential nutrients hypothesised to be involved in reducing the likelihood of autism such as long-chain poly-unsaturated fatty acids (LCPUFA) [22,23]”. Again the authors should make it clear that the main exposure problem is with methyl mercury.

This has been corrected. [Line 85-87]

 

7. Lines 185-186 It states “The mean Hg in maternal whole blood ranged from 0.64 [39] to 3.71 μg/L, and maternal hair sample concentrations from 0.3 to 5.7 μg/g”. This may be the smoking gun for the discrepancy between this and other literature. The values selected are really quite low. Hair values of 20 to 50 ug/g should have been included and of course there is a safety factor involved when scaling down to the low values considered in this analysis. The WHO Safe limit for children and women of childbearing age is 2 ppm. USEPA safe levels are even less.

We have found studies with higher levels of Hg, but none were identified in our literature search which contained both our target neurodevelopmental outcomes and age range of 0 to 5 years.

A paragraph discussing the possibility of a neurotoxic effect above a higher threshold has been added to the discussion. However, we could not find many guidelines relevant to both (a) circulating Hg levels rather than Hg intake and (b)  maternal Hg and the safety of the developing fetus. 

The guidelines we found reported ‘safe limits’ for pregnant women that overlapped (FEA: 2.0 µg/L, Mahaffey et al: 3.5 µg/L) or were close to (EPA: 5.8 µg/L) the range reported by studies in this review. We found further guidelines with much higher values (e.g. Brodkin et al, 2007: 10-40 µg/L) but these appeared concerned with mercury poisoning rather than specific to pregnant women. If guidelines with higher values have been overlooked, we would be glad to add them and modify our conclusions further.

8. Lines 435-436 gives the statement “The evidence for an association between prenatal mercury concentrations and neurodevelopmental functioning below the age of 5 years old is weak”. This reviewer suggests adding “in the range of concentrations considered in this analysis”. The bottom line is they looked at levels near the safe level and so seeing a weak association would be expected.

The qualifier “At the levels of mercury recorded in studies in this review,” has been added to the conclusion. [Line 464]

 

 

8. Two confounding factors which were not addressed in this review was whether or not the mothers had mercury amalgam fillings in their teeth or if the young children were given vaccines containing mercury. These problems have not been approached responsibly as many researchers have had vested interests in the outcomes. In vaccines the mercury is in the form of trimethyl mercury called Thermarasil and the chemical methods used were not reliable.

We agree that it would be interesting to compare studies which measured mercury exposure from amalgam fillings or vaccines with those from this review. For this review we did not consider either to be a key confounder of the relationship between maternal Hg and neurodevelopment:

Mercury exposure from infant vaccination would occur after maternal Hg was measured. Vaccine Hg could at best be a competing exposure on neurodevelopment, but without a way to affect maternal Hg it could not confound the effect of interest.

Maternal dental amalgams with Hg could cause higher levels of circulating Hg to be recorded, and this could increase fetal Hg exposure and lead to poorer performance on neurodevelopmental assessments. However, the potential effect of dental amalgam on neurodevelopment appears to be entirely through Hg toxicity, so we expect it to be a source of maternal Hg levels rather than a confounder on maternal Hg and neurodevelopment.

Reviewer 3 Report

The review in its logical structure is done well. There are just a few things that are not clear to me.

It is not very clear to me why studies with children over 5 years old were excluded (as written in the discussion). I ask you to make this choice more explicit.

Another thing I ask is to specify in a somewhat more systematic way the strength of the effect of the various studies. Many studies that do not find the association are in fact "under-powered". So perhaps for each study it would have been convenient to also report the strength of the effect in the final table as well as the sample size.

In the conclusions the authors do not find a clear association between mercury and problems in cognitive development. It would be interesting to apply a Bayesian-type approach to studies that do not find associations between mercury and cognitive development, if it were possible and compatible with the experimental design. The Bayesian approach a priori assigns a probability to the hypotheses and can actually calculate whether the absence of a relationship is more likely.

In the conclusions the authors affirm that for the works considered reliable there are no certain evidences of the relationship mercury neurobehavioral effects or cognitive development. Especially when mercury is consumed in the diet (mercury is mainly present in fish) but the diet itself can mitigate the effect of mercury, so generally for a number of toxic substances. It is no coincidence that these manifestations are observed above all in poor countries.
In other words, it is really complex to correlate neurobehavioral manifestations towards subchronic exposures for single compounds.

Author Response

Thank you for considering our paper. We have made adjustments which we hope meet your concerns and suggestions.

1. It is not very clear to me why studies with children over 5 years old were excluded (as written in the discussion). I ask you to make this choice more explicit.

The age range has been made more explicit in the title, abstract, and method sections of the paper, and a brief justification added to the methods. [Line 20 , Table 1, Lines 148-152]

2. Another thing I ask is to specify in a somewhat more systematic way the strength of the effect of the various studies. Many studies that do not find the association are in fact "under-powered". So perhaps for each study it would have been convenient to also report the strength of the effect in the final table as well as the sample size.

Results from high quality studies have been moved into the main text under each neurodevelopmental subcategory, which we hope makes it easier for the reader to assess themselves the strength of evidence. [Tables 4-9]

It is possible that more studies are underpowered than the few we identified, but it is difficult to clearly determine this because any estimate which overlaps with the null could be argued to be underpowered. The sample size and strength of the effect have both been included in tables 4-9.

 

3. In the conclusions the authors do not find a clear association between mercury and problems in cognitive development. It would be interesting to apply a Bayesian-type approach to studies that do not find associations between mercury and cognitive development, if it were possible and compatible with the experimental design. The Bayesian approach a priori assigns a probability to the hypotheses and can actually calculate whether the absence of a relationship is more likely.

We agree this would be an interesting approach. The original goal of this project was to meta-analyse results, which would have given a clearer summary of the evidence. Unfortunately the methodological differences between studies were too great– even when restricting to only those studies which use a common measure such as BSID-II MDI. This was due to differences in mercury measurement methods and exposure/outcome timings.

 

4. In the conclusions the authors affirm that for the works considered reliable there are no certain evidences of the relationship mercury neurobehavioral effects or cognitive development. Especially when mercury is consumed in the diet (mercury is mainly present in fish) but the diet itself can mitigate the effect of mercury, so generally for a number of toxic substances. It is no coincidence that these manifestations are observed above all in poor countries.

In other words, it is really complex to correlate neurobehavioral manifestations towards subchronic exposures for single compounds.

We agree that these are the key conclusions from this review. The conclusion has been modified to reflect this. [Lines 599-604]

Reviewer 4 Report

Dear authors,

This is a well-written manuscript.  My only suggestion is for the results section, I believe that the part 8 table in the supplementary data should be included in the manuscript, maybe broken down into 6 tables (if needed) to go with text.  Also please check the year of the references, for example in supplementary data part 8, you have Kim Y et al. 2008 when it should be 2018.

Author Response

Thank you for reviewing the paper and for your positive feedback. We have made adjustments which we hope meet your suggestions.

1. This is a well-written manuscript. My only suggestion is for the results section, I believe that the part 8 table in the supplementary data should be included in the manuscript, maybe broken down into 6 tables (if needed) to go with text.

The results from part 8 have been moved into the main text, divided by subcategory as suggested. [Tables 4-9]

The large number of columns makes this difficult to fit into the journal template. We have requested help from the editor to improve the formatting of this.

 

2. Also please check the year of the references, for example in supplementary data part 8, you have Kim Y et al. 2008 when it should be 2018.

This has been amended and all other references checked.

Round 2

Reviewer 1 Report

Thanks for your response. I see. This is a systematic review article. I accept your explanation. For the better quality of the manuscript, I still suggest to do several minor changes. 
1. Please change BSID—III as Bayley-III in the whole manuscript.
2. Based on the current abstract, the audiences can not get the informative results from this article. It is recommended to rewrite it again. I hope your article to obtain the high cited number. According to the current abstract, I am afraid.
3. The conclusion is also needed to be rewritten and changed.
4. Can you encourage the future studies or work based on your review article.

Author Response

1. Please change BSID—III as Bayley-III in the whole manuscript. 

Bayley-III is now used in all the manuscript.

2. Based on the current abstract, the audiences can not get the informative results from this article. It is recommended to rewrite it again. I hope your article to obtain the high cited number. According to the current abstract, I am afraid.

The abstract results and conclusions have been rewritten to make our primary findings clearer.

3. The conclusion is also needed to be rewritten and changed.

Unfortunately, it is not clear which part of the conclusion needs to be written, or how it should be changed. Following changes to the conclusion based on the previous round of peer review, we believe the conclusion to be adequate: it summarises our primary finding, affirms the lack of pattern by study methodology, and gives a biological and clinical interpretation, in as few sentences as possible.

 
4. Can you encourage the future studies or work based on your review article.

This has been added to the conclusion.