Finally into the swing of things for the new year and maybe it’s time to look a little more seriously at neurotoxicity. This is a big topic and obviously one of pretty intense interest in paeds anaesthesia. So this is the first in a series where I’ll try to work through at least some of the different twists in the story. Not every single twist. Just some of them. The aim is to churn these out about a week apart and before you know it we’ll have a whole suite.
Most of us in paediatric anaesthesia are probably in it, at least in part, because we quite like the idea of making things better for kids. We stop pain, we make operations OK and we sometimes blow bubbles at work.
Actually some of us are probably in it at least a bit for the bubbles.
And when you consider that it wasn’t until about 1987 that we settled the question of whether infants even needed a comprehensive approach to anaesthesia and analgesia because ‘maybe pain isn’t really a thing for them’, we’ve come a fair way.
So the emergence of evidence that maybe the stuff we use for the core of our job might damage neurons was, and is, a big deal. It became an urgent question to explore and here we are, approaching 19 years later, still trying to understand it.
And it all came about from someone reversing an old saying and deciding that maybe you should work with animals and kids.
The Rats Speak
Back in 1999, Ikonomidou and crew went and talked to the animals. I should be clear that when I say “animals” I mean specifically lab-living rodents and by “talked” I mean “they injected 7 day old rats with 3 doses of either a vehicle (no, not a tiny car, but more like a rat placebo) or the NMDA antagonist dizocilpine and later sacrificed those rats so they could examine their brain tissue”.
What they report in their tiny-font article was a little confronting. When they looked at the brains of those on the receiving end of the vehicle they saw a few degenerating neurons here and there in the forebrain.
When they looked all over the place in those getting the dizocilpine (let’s use the catchier version, MK801) increased the density of degenerating neurons from 3 all the way to 39 times depending on where they looked. Reported another way, the density of degenerating neurons was 0.2-1.55% of the total neuronal density for the vehicle group and as high as 15-26% of the total density in the MK801 group.
What’s more, degenerating neurons looked just like neurons look when apoptosis has kicked in. They didn’t stop at a few brief answers on that either. They checked brains at 4, 8, 12, 16, 24 and 48 hours after the MK801 and the apoptosis at 4 hours got progressively worse through the 12-24 hour range.
But what if it’s not the NMDA blocking bit? Well they cross-checked with phencyclidine, ketamine and carboxypiperazin-4-yl-propyl-1-phosphonic acid (CPP) and the cell degeneration looked equally startling. Use the less active enantiomer of MK801 and the neurodegeneration was less too.
But wait, there’s more…
They just hadn’t sacrificed enough rats to the research gods so they tried out the MK801 (or the vehicle) on postnatal days 0, 3, 7, 14 and 21 and also treated pregnant rats on embryonic days 17, 19 or 21 and checked out the foetal rat pups.
The rats that got the drug on postnatal days 0 through 14 still showed the apoptosis (particularly for days 3-7, though days 0-3 weren’t great either). They also showed that the response in different brain regions was different.
Then, given they just hadn’t tested things on enough rats, they tried out the effect of an antagonist of non-NMDA glutamate receptors, scopolamine as an anticholinergic at muscarinic receptors and haloperidol as an antagonist at dopamine receptors. Plus they blocked calcium influx with calcium-channel blockers. None of those agents lead to apoptosis.
Damn. So many drugs. So many rats. And at the end of it, pretty fair worries about what happened when you blocked receptors that we might just want to block during an anaesthetic. And it wasn’t too long before there were similar findings in animals given agents acting as agonists at GABA-A receptors.
But cells don’t equal later function necessarily, right? I mean the young neurons around here and there can pick up the slack?
Will they ever learn?
If there was one reassuring thing it was that cells are one thing and function is another thing again, right?
Well in 2003 Jevtovic-Todorovic et al reported findings after exposing the developing rat brain to isoflurane, midazolam and nitrous oxide. The rats were 7 days old again and the exposure to anaesthesia was 6 hours and then they got back to the looking at the brain bit at days 29-33.
There was a key difference this time around because rats were also made to jump through scientific hoops. There were a lot of paces too. Responses to known drug treatments, startle tests, locomotor activities, sensorimotor tests and spatial reference and working memory.
Oh, and water navigation. I guess that test arose from the need to know how they went deserting ships back in the day.
Oh and I also forgot to mention that along the way they did arterial blood gas analysis to check the effect of anaesthesia on metabolism and stress responses.
This group managed to find evidence of widespread apoptosis with isoflurane. How widespread? Well here’s just one of the ways they showed it in the paper.
But they also found functional impairments in spatial learning and memory. In particular these rats took longer to learn things when testing spatial learning. They still got there. But they were slow on the uptake.
Perhaps the key thing to note though was that these impairments were not the sort of things you’d pick up just from hanging out with your pet rat. You have to look carefully.
The problem with rats of course is that they are, well, rats. Skipping on to 2010 and Brambrink and crew came up with a protocol to test things out on macaques (and no I could not bring myself to find macaque pictures).
Isoflurane anaesthesia was maintained for 5 hours then 3 hours after the anaesthetic, which really isn’t long, they got to the sacrificing bit and found the rate of apoptosis was 13 times above that of controls. Different brain regions showed different patterns of neurodegeneration again.
The rats and monkeys were both trying to tell us something.
That isn’t the only work on non-human primates of course and as a slightly different example, in 2015 Lisheng Zhou et al did work where they exposed Cynomolgus monkeys to sevoflurane for 5 hours. Sevoflurane didn’t alter behaviour in a holding cage test. It didn’t affect learning. It didn’t affect memory.
Wait, why doesn’t it fit? There’s always variability and I guess it’s worth remembering that not all things were universally the same.
Which leaves us where?
Absolutely unsurprisingly, no one was excited to think the drugs might work but also unwork the brain. The animal evidence just kept suggesting that most of the agents used to tackle the awareness part of the anaesthetic had implications for developing neurones.
And of course there are plenty of reasons you can’t just shut the whole anaesthesia thing down. Not just dodgy reasons you come up with to justify your continuing use of a thing you just don’t want to face giving up. Real reasons for cautious interpretation of the animal research, which you can pretty much break up into:
What if it’s not the drugs, it’s the physiology?
This argument arose from the observation that most of the time we provide higher level anaesthesia care than was offered to those early rodents. They just didn’t work hard enough at oxygenation, ventilation and haemodynamics and maybe the wacky physiology caused the damage.
And there’s something to this. If you don’t know the haemodynamics and gas exchange that go with the findings, its a bit tough to figure out whether the rat pups could have been compromised just because the physiology wasn’t monitored and acted on where needed.
Of course in that macaque study they clinically monitored depth of anaesthesia, intubated with a semirigid fibreoptic scope, mechanically ventilated and maintained with continuous end-tidal capnography, gas analysis, ECG, pulse oximeter noninvasive blood pressure (although only 15 minutely) and allowed for temperature monitoring. Oh and they got a gas via direct sampling from the ventricle.
Not that bad then. Or at least should probably not assume that physiology is the whole story.
We haven’t seen the evidence in people and we’ve been doing it for ages
This one was true but at the same time no one was really looking either. And in the context of differential effects on different brain sections, and those learning changes in very specific things in the rodents, it’s not as easy as delivering a single version of a functional test. You’d need to check multiple ways.
That is not anaesthesia, that is crazy overdosing
This one is part of the broader “how do you extrapolate?” discussion. In those initial rodent studies they aimed for 8 hours of NMDA antagonism during peak synaptogenesis (a period of 14 days) for an animal that has a lifespan of about 3 years. On rough calculations that’s about 2.4% of the total period of peak synaptogenesis. In fact 8 hours would be a very long surgery which means a big exposure to the agent. You could say the same for 5 hours in the monkey research team or the up to 24 hour long infusions of drugs like ketamine described in other studies.
If you take the period of peak synaptogenesis for humans, also referred to in one of these papers as the brain growth spurt, as the last month in the womb and the 6 months after, and then figure out what proportion of that time a 6 hour anaesthetic comes to, you’re somewhere like 0.12% of the period of peak synapse formation. And most cases are nowhere near 6 hours. This was enough to make lots of people pretty relaxed about it all. Exposure for neonates just isn’t that long in most cases.
And that’s also assuming you can really correlate the maximal brain growth phases across species. Which brings me to…
How do you match species?
Plenty of people weren’t entirely sure about the animal models. How much can you trust that the estimates from the animal researchers regarding when the peak period of synapse formation is and whether that truly correlates with a particular epoch in kids? And if you have a longer period of peak synapse formation, like when you compare the development of the human brain versus the rodent brain, are the key times the same? And how do you weigh up the larger number of connections in humans, where maybe there is more scope for adaptation thanks to neuroplasticity?
I remember hearing plenty of dismissive comments about the animal models or at least uncertainty.
They forgot about the alternative
I don’t mean the alternative for a lab rat which probably still involves an early sign off and a microscope. I’m talking about the alternative to anaesthesia.
The alternative to anaesthesia for surgery is no anaesthesia. And the experience of surgery without anaesthesia has its own effects on neurodevelopment. So at least some comments around the time were along the lines of ‘OK, but what’s the other option? There is no other option. All we could do anyway is try and minimise it.’
So what were the animals telling researchers? Well, in a lab setting, and with different types of animals, both agents that target NMDA receptors and agents that target gabaminergic pathways are associated with scary looking neurodegeneration that looks just like apoptosis. Different areas are affected differently. And there is at least some evidence that those exposed animals have issues with learning at a later stage.
And all the attempts to provide context or other explanations can’t really divert you from thinking that more questions needed to be asked. It was no longer OK to say ‘we’ve done this for ages and kids are fine’. It was obvious that it was time to start really actively looking. But that bit can be for the next instalment, in which we’ll leave the rats behind. Or leave them to drink Starbucks maybe.
That image of Starbucks rat came from unsplash.com which has lots of images available to share. This one was put up by Mert Guller and is unchanged here.
This post is not an attempt to cover every single bit of animal research that is part of the literature around neurotoxicity. These are just a small selection of prominent papers, particularly the Ikonomidou work that really kicked things off.
If you have other key points about the animal literature relevant to neurotoxicity I am all ears and would welcome comments and corrections. Although I went through these papers in detail, bits were fairly heavy going so if you spot something you think deserves an alternate take, I suspect everyone (well at least me) would learn from it.
And this is the bit where I insert the standard mention that sharing things is always appreciated and if you like the things around here you could consider signing up to get an email whenever things drop. Look around. The sign up thing will be here.
Anyway, here are some links so you can go back to the source literature mentioned in this post:
Jevtovic-Todorovic V, Hartman RE, Izumi Y, et al. Early Exposure to Common Anesthetic Agents Causes Widespread Neurodegeneration in the Developing Rat Brain and Persistent Learning Deficits. J Neurosci. 2003;23:876-82.
Zhou L, Wang Z, Zhou H, et al. Neonatal exposure to sevoflurane may not cause learning and memory deficits and behavioural abnormality in the childhood of Cynomolgus monkeys. Scientific Reports. 2015;5;Article number: 11145. doi:10.1038/srep11145.
Now you have come a long way and if you’d like a diversion then you could do worse than watching this blue marble do its thing.