Zinc, lead, cadmium, uranium and more scrutinized in a new study.

It has always amazed me that our bodies require these tiny amounts of incredibly rare substances to function. Like, sure, we need oxygen – we need water.  But we also need molybdenum – which makes up just 1.2 parts per million of the Earths crust.

Without adequate molybdenum intake, we develop seizures, developmental delays, death. Fortunately, we need so little molybdenum that true molybdenum deficiency is incredibly rare – only seen in people on total parenteral nutrition without supplementation or certain rare genetic conditions. But still – molybdenum is necessary for life.

Many metals are. This chart colors the essential minerals on the periodic table. You can see that to stay alive we humans need not only things like sodium, but selenium, bromine, zinc, copper, cobalt.

And some metals are very clearly not essential – we can all do without lead and mercury, and probably should.

But just because something is essential for life does not mean that more is better. The dose is the poison, as they say. And this week, we explore whether metals – even essential metals – might be adversely affecting our brains.

It’s not a stretch to think that metal intake may have weird effects on our nervous system. Lead exposure, primarily due to leaded gasoline, has been blamed for an average reduction of about 3 points in our national IQ, for example. But not all metals are created equal. Researchers set out to find out which might be more strongly associated with performance on cognitive tests and dementia and report their results in this study in JAMA Network Open.

To do this, they leveraged the “MESA” cohort study. This is a longitudinal study of a relatively diverse group of 6300 adults who were enrolled from 2000 to 2002 around the United States. At enrollment, they gave a urine sample and took a variety of cognitive tests. Important for this study was the digit substitution test, where participants are provided a code and need to replace a list of numbers with symbols as per that code. Performance on this test worsens with age, depression, and cognitive impairment.

Participants were followed for more than a decade and over that time 559 (about 9%) were diagnosed with dementia.

Those baseline urine samples were assayed for a variety of metals, some essential, some very much not, as you can see here.

Now, I have to put my kidney doctor hat on for a second and talk about urine measurement… of anything. The problem with urine is that the concentration can change a lot – by more than 10-fold in fact – based on how much water you have drunk recently. Researchers have to correct for this, and in the case of this study they do what a lot of researchers do, they divide the measured concentration by urine creatinine. 

This introduces a bit of a problem. Take two people with exactly the same kidney function, who drank exactly the same water, whose urine is exactly the same concentration.  The person with more muscle mass will have more creatinine in that urine sample – since creatinine is a byproduct of muscle metabolism.  Now… since people with more muscle mass are generally healthier, when you divide your metal concentration by urine creatinine, you get a lower number which might lead you to believe that lower levels of the metal in the urine are protective.  When in fact what you’re seeing is that higher levels of creatinine are protective.  I see this issue all the time and it will always color results of studies like this. I’ll mention some ways around it in a minute.

OK, I am doffing my kidney doctor hat now to show you the results.

The researchers first looked at the relationship between metal concentrations in the urine and performance on cognitive tests. The results were fairly equivocal, save for that digit substitution test which I’m showing you here.

Even these results don’t ring major alarm bells for me.  What you’re seeing here is the change in scores on the digit substitution test for each 25-percentile increase in urinary metal level – a pretty big change. And yet you see really minor changes in the performance on the test. The digit substitution test is not an IQ test, but to give you a feeling for the magnitude of this change, if we looked at copper level, moving from the 25th to the 50th percentile would be associated with a loss of 9/10ths of an IQ point.

You see two colors on this graph by the way. That’s because the researchers stratified their findings based on whether the individual carried the ApoE4 gene allele – which is a risk factor for the development of dementia. There are reasons to believe neurotoxic metals might be worse in this population, and I suppose you do see generally more adverse effects on scores in the red lines compared to the blue lines – but still, we’re not talking about a huge effect size here.

Let’s look at the relationship between these metals and the development of dementia itself – a clearly more important outcome than how well you can replace numeric digits with symbols.

I’ll highlight of few of the results that are particularly telling.

First, the non-essential mineral cadmium – which displays the type of relationship we would expect if the metal was neurotoxic: a clear, roughly linear increase in risk of dementia as urinary concentration increases.

We see roughly similar patterns with the non-essential minerals tungsten and uranium, and the essential mineral zinc (beloved of respiratory-virus avoiders everywhere).

But it is very much not what we see for all metals. Strangest of all, look at lead which shows basically no relationship with dementia.

This concerns me a bit. Earlier, I discussed the issue of measuring stuff in urine and how standardizing levels to urine creatinine introduces a bias due to muscle mass. One way around this is to standardize urine levels to some other marker of urine dilution – like osmolality.  But more fundamental than that, I like to see positive and negative controls in a study like this.

Like – lead would strike me as a good positive control here. If the experimental framework were valid, I would think we’d see a relationship between lead level and dementia. For a negative control? Well, something we are quite sure is not neurotoxic – something like sulfur maybe that is relatively ubiquitous, used in a variety of biological processes, and efficiently eliminated. We don’t have that in this study.

The authors close their case by creating a model that combines all the metal levels – asking the question of whether higher levels of metals in the urine in general worsens cognitive scores. And they find that the relationship exists, as you can see here, both in carriers and non-carriers of ApoE4. But, to me, this is even more argument for the creatinine problem. If it’s not a specific metal, but just the sort of general concentration of all metals, the risk of confounding by muscle mass is even higher.

So should we worry about ingesting metals? I suppose the answer is… kind of?

I am sure we should be avoiding lead, despite the results of this study. Probably best to stay away from uranium too.

As for the essential metals? I’m sure there is some toxic dose – there’s a toxic dose of everything at some point. But I don’t see evidence in this study to make me worry that a significant chunk of the population is anywhere close to that.

A version of this commentary first appeared on Medscape.com.