Time to Re-evaluate the Guideline Value for Manganese in Drinking Water?

This 2007 review by Karin Ljung and Marie Vahter of the Institute of Environmental Medicine, Karolinska Institutet, examined the scientific basis for the WHO guideline value for manganese in drinking water and concluded that the derivation relied on debatable assumptions, that the NOAEL used was not scientifically well-supported, and that the guideline provided insufficient protection for infants. The authors argued that preparation of infant formula with water at the WHO guideline concentration would cause the formula to exceed the European Commission’s maximum manganese value for infant formula, and that the physiological differences between infants and adults made extrapolation of adult-derived limits inappropriate.

Key numbers

Manganese occurrence in drinking water:

WHO guideline and its derivation:

• WHO guideline at time of review (3rd edition, 2006): 400 µg/L (health-based); prior 2nd edition value was 500 µg/L.
• Derivation used a NOAEL of 11 mg manganese/day, divided by an uncertainty factor of 3 to allow for higher bioavailability from water versus food, giving a TDI of 60 µg/kg body weight/day for a 60 kg adult; 20% allocation to drinking water and 2 L/day assumption yielded 400 µg/L.
• The 11 mg/day NOAEL traced through WHO and IOM documents back to a single Canadian study of 100 women (Gibson and Scythes 1982), where calculated (not analyzed) intakes reached a maximum of 10.8 mg/day. Only 10% of women ingested more than 5 mg/day; the analyzed mean was 2.4 mg/day. The authors found no basis in that study for designating 11 mg/day as a NOAEL.
• Table 1 of the paper summarizes daily manganese intakes across countries and age groups from multiple studies: children on omnivorous diets 1.3 to 3.6 mg/day (mean approximately 2 mg/day); adults on omnivorous diets 1.5 to 3.9 mg/day (mean approximately 2.7 mg/day); vegetarian diets somewhat higher. None of the studies found intakes as high as 11 mg/day.

Manganese in infant formula (reconstituted):

The authors report that most standard and allergenic formulas have manganese concentrations of approximately 400 µg/L and that formula contains approximately 100-fold higher manganese concentrations than breast milk, before any contribution from preparation water.

Manganese in breast milk:

Infant formula regulatory values and LOAEL:

• European Commission SCF (2003) maximum manganese for infant and follow-on formula: 100 µg/100 kcal, equivalent to approximately 650 µg/L for a reference infant consuming 0.78 L/day at 500 kcal.
• When standard formula containing an average of 325 µg/L is mixed with water at the WHO guideline value of 400 µg/L, the SCF maximum is exceeded (Sievers 2005, as cited in the paper).
• IOM adequate intake (AI) for manganese: 3 µg/day for infants under 7 months (based on 0.78 L/day breast milk at 3.5 µg/L); 600 µg/day for infants 7 to 12 months.
• IOM tolerable upper limit (UL) for adults: 11 mg/day; UL for infants under 1 year: not established, due to insufficient data.
• LOAEL identified by Velazquez and Du (1994) from the Kondakis et al. (1989) study of adults over 50 years consuming water at 1,800 to 2,300 µg/L for more than 10 years: 2 mg/L. US EPA converted this to 4.2 mg/day for a 70 kg adult consuming 2 L/day. The SCF cited this as 4.2 mg/L, which the review authors identify as a units error making the applied LOAEL more than twice the actual water concentration from the Kondakis study. The body-weight-scaled LOAEL for a 5 kg infant is approximately 300 µg/day (385 µg/L at 0.78 L/day, corresponding to 60 µg/100 kcal), substantially below the formula maximum of 100 µg/100 kcal.

Table 2 of the paper (manganese guideline and calculated values by age group):

Infant physiology and manganese:

• Infants absorb approximately 20% of dietary manganese from formula, versus 1 to 5% in adults (Dorner et al. 1989, as cited).
• Cord blood manganese concentrations are approximately 150% higher than maternal serum (Rossipal et al. 2000; Krachler et al. 1999, as cited).
• Geometric mean cord blood manganese: 39 µg/L (range 15 to 93) in one study (n = 222 newborns); 34 µg/L (range 17 to 90) in a second (n = 87 newborns) (Takser et al. 2003, 2004, as cited).
• Reference blood manganese in healthy adults: 5.0 to 12.8 µg/L; median 7.6 µg/L (Goulle et al. 2005, as cited).

Neurotoxicity studies cited in the review:

• He et al. (1994) and Zhang et al. (1995): 92 matched-pair children aged 11 to 13 years; exposed group drank water at 240 to 350 µg/L vs control group at 30 to 40 µg/L; exposed children had lower neurobehavioral scores and lower mathematics and language scores.
• Woolf et al. (2002): 10-year-old boy with 5 years’ exposure to water at 1,200 µg/L; below-average memory, elevated whole blood/urine/hair manganese.
• Wasserman et al. (2006): 142 children, 10 years old; dose-response association between well water manganese (average 800 µg/L; range 4 to 4,000 µg/L) and performance and verbal ability test scores.
• Bouchard et al. (2007): 46 children; 28 in homes with average tap water at 600 µg/L vs remaining children at average 160 µg/L; hyperactivity and oppositional behavior differences; hair manganese correlated with water concentration.
• Takser et al. (2003): negative associations between cord blood manganese and attention, nonverbal memory, and hand skills at age 3; no significant associations at age 6.
• Golub et al. (2005): infant rhesus monkeys fed soy formula (300 µg/L) showed reduced play and shorter wake cycles compared with cow milk controls (50 µg/L).

Study design

This is a systematic narrative review, not a primary measurement study. The authors traced the derivation chain of the WHO manganese guideline through the WHO background document, IOM dietary reference intake reports, and original research articles. They also searched for published studies on manganese occurrence concentrations and neurotoxic health effects in children and infants. No original laboratory analysis was performed. Data synthesis compared guideline-derivation assumptions against the broader literature on dietary manganese intakes, infant physiology, and neurodevelopmental outcomes. Sample sizes and sampling details are those of the individual cited studies, not of this review.

Methods (brief)

No primary analytical measurements were performed. Data sources were the WHO 2004 background document for the manganese drinking-water guideline, IOM dietary reference intake reports, and primary research articles retrieved from citation tracing and literature search. Manganese occurrence values for Swedish groundwater are cited from the Geological Survey of Sweden (unpublished at time of publication); US values are from the US EPA 2003 health effects support document. Formula manganese concentrations are from Hozyasz and Ruszczynska (2004) and from direct manufacturer data reported in a companion report (Ljung et al. 2007, IMM-report 1/2007). Analytical methods for the cited studies included inductively coupled plasma mass spectrometry (ICP-MS; Krachler et al. 2000) and conventional dietary balance techniques (Dorner et al. 1989); methods for other cited studies are as described in those original papers.

No limit of detection, limit of quantification, or certified reference material recovery is reported for this review article, as none is applicable to a secondary analysis.

Evidence fitness

This source supports: assessment of manganese occurrence in groundwater and drinking water (occurrence data cited from multiple jurisdictions); characterization of the derivation pathway and identified weaknesses of the WHO 400 µg/L guideline; estimation of manganese intake in infants from formula and breast milk; review of neurotoxicity evidence in children at drinking water concentrations below the WHO guideline; and identification of the regulatory gap between the formula maximum (100 µg/100 kcal) and the LOAEL scaled to infant body weight.

This source does not support: certification limit-setting for lead, cadmium, arsenic, or mercury; primary analytical measurement of manganese in any food or water; brand-level contamination assessment; or causal attribution of specific neurodevelopmental outcomes to any defined manganese exposure level (the authors explicitly state that causality remains uncertain and further research is needed).

Limitations

The review is a narrative rather than systematic meta-analysis. No pooled quantitative synthesis of health-effect study results was performed. Neurotoxicity effect estimates from cited studies are not independently re-analyzed. The Swedish groundwater data cited (SGU) were unpublished at the time of publication and cannot be independently verified from the paper alone. The derivation critique applies to the WHO 2006 third edition and the 2004 background document; subsequent WHO guidance has revised the derivation and guideline value. The infant LOAEL calculation uses body-weight scaling from an adult study population (Kondakis 1989) in adults over 50 years, not in infants or children, a limitation the authors acknowledge. Manganese is an essential nutrient as well as a potential neurotoxicant at high exposure; the review does not examine deficiency risk at very low formula concentrations.

Regulatory context

At the time of publication, the WHO guideline for manganese in drinking water was 400 µg/L (3rd edition, 2006); the prior 2nd edition value was 500 µg/L. The US EPA secondary maximum contaminant level (aesthetic) was 50 µg/L; the health reference level was 300 µg/L. The European Commission SCF maximum for manganese in infant formula was 100 µg/100 kcal (approximately 650 µg/L reconstituted). The IOM had not established a tolerable upper limit for infants under 1 year. The WHO subsequently revised the guideline; for the current applicable values see Manganese and Drinking water in the food supply chain.

Related evidence

• Manganese
• Infant Formula, Powder (Non-Soy)
• Water
• Drinking water in the food supply chain

Update history

No substantive edit history is available in this build. The full commit record is available in git.