Help me go beyond high school chem class, LD50-wise.

August 8, 2007 6:01 AM

Calling all science types: Help me get a more sophisticated grasp of the "LD50" concept used to test the toxicity of substances.

I understand the basic idea of LD50 (the dose of a substance per kilo of body weight that will be lethal to half of the test subjects), but have some questions about specifics.

Namely: all of the LD50 experiments are conducted on animals- mostly rats, it seems. How to researchers extrapolate their data from these experiments to apply to humans? Given what I presume are the differences between human and rat metabolism, it seems like a very speculative and inaccurate way to gauge the threshold a "lethal dose" of a drug in humans, and I wonder how and if researchers/toxicologists etc. account for this leap.

As always, thanks for any and all help.

posted by foxy_hedgehog to science & nature (17 comments total)

Can't be known with certainty (short of, you know, actually testing it on people). Really specific compounds (venoms, neurotoxins etc.) will have different toxicity in different species (the antibody TGN1412 is an obvious example), but I'd expect that for more broad-acting substances (arsenic, cyanide) the human LD50 is very similar to the rodent LD50.
posted by kisch mokusch at 6:37 AM on August 8, 2007

They don't, AFAIK. As you're probably aware, LD₅₀ is typically reported as mg/kg, so if 10mg of a substance is lethal to a 1kg rat, our "best guess" is that it would take 800mg to be lethal to a 80kg human.

But you're right in that different metabolisms and different biochemistries, it is very speculative to extrapolate from other species to humans. And in fact vast differences are sometimes seen in LD₅₀s measured for the same substance in different species. The LD₅₀ for theobromine (the ingredient in chocolate which is harmful to dogs) in dogs would be far too low if applied to humans.

That's why LD₅₀ is always reported with the species the test was done in. A proper LD₅₀ isn't just "10 mg/kg," it's "10 mg/kg (rat)." If you have LD₅₀s for the same substance from multiple species, this might inform your estimate of what the LD₅₀ in humans is: it's probably closer to the LD₅₀ in chimpanzees than the LD₅₀ in mice, for example.
posted by DevilsAdvocate at 6:44 AM on August 8, 2007 [1 favorite has favorites]

The rigorous scientific way of doing this is to figure out the amount that causes harm in rats and then divide that by 1,000 to create an official safe exposure for humans. That's the first cut. From there it can be refined more as needed.

More generally applicable than LD-50, though, is NOAEL (No Observable Adverse Effect Level).

Note that this stuff is very controversial. Scientists and industry have been arguing about the applicability of rodent data on BPA to human for many years. The product continues to be used all over the place, despite numerous studies showing serious adverse effects in rodents.
posted by alms at 6:47 AM on August 8, 2007 [1 favorite has favorites]

Given what I presume are the differences between human and rat metabolism

Mammals actually have roughly similar metabolism for a lot of substances. The LD50 (human) is unlikely to be equal to the LD50 (rat), but it is likely to have the same order of magnitude. After dividing by a safety factor of 1000-10,000 you can have some certainty that humans will not be killed by it. But toxicologists will need to do other research on the substance, not least of which because they are interested in other harmful effects besides death.

A Canadian Centre for Occupational Health and Safety factsheet may be informative.
posted by grouse at 6:48 AM on August 8, 2007 [1 favorite has favorites]

So interesting. So really the LD50 is about getting a thumbnail sketch to gauge what might be a reasonable and non-harmful dose in humans.

Grouse's site says:
Where can I find LD50 and LC50 values?

The largest, single collection of LD50 and LC50 values is in the database Registry of Toxic Effects of Chemical Substances (RTECS) that is available by subscription on CD-ROM and on the Internet. Two other databases available from CCOHS, CHEMINFO and the Hazardous Substances Data Bank® (HSDB). Both of these are on the CHEMpendium CD-ROM; CHEMINFO is also accessible on the Internet.

Is this kind of info only available via the pricey databases in order to limit access to the scientific community? Or is it also in, say, the PDR?
posted by foxy_hedgehog at 6:56 AM on August 8, 2007

The Merck Index also reports LD50 for many compounds. Older copies are inexpensive.
posted by exogenous at 7:06 AM on August 8, 2007

So really the LD50 is about getting a thumbnail sketch to gauge what might be a reasonable and non-harmful dose in humans.

I think you've got it.

Is this kind of info only available via the pricey databases in order to limit access to the scientific community?

No one is trying to limit access to the scientific community, they're just trying to make money. LD50 information should be in Material Data Safety Sheets, and you can find many MSDSs on the web, but many of the sites require registration.
posted by grouse at 7:11 AM on August 8, 2007 [1 favorite has favorites]

If you have a college or university check their libraries. Our's has quite a selection of Merck and chemical reference books that have just this sort of thing in them.
posted by LunaticFringe at 8:22 AM on August 8, 2007

So really the LD50 is about getting a thumbnail sketch to gauge what might be a reasonable and non-harmful dose in humans.

LD50 is the smallest median lethal dose that has been reported in any experimental animal by either oral administration or skin application. I would think it is about getting a thumbnail sketch to gauge what might kill you. Do not assume that a dose smaller than the LD50 would be non-harmful. Depending on the steepness of the dose-response curve (the relationship between the smallest dose that will kill any animals and the largest dose that some of the animals will survive) the dangerous dose for humans may be one to ten percent less than the indicated lethal dose.
posted by weapons-grade pandemonium at 8:57 AM on August 8, 2007 [1 favorite has favorites]

In many cases, drug LD50 data for humans is obtained by case reports of overdoses; while these are obviously not done under controlled circumstances and are typically a pretty small set of data, they do provide some useful data.
posted by TedW at 9:11 AM on August 8, 2007 [1 favorite has favorites]

Great info. Even more interesting and complicated than I thought, which is usually the case. Thanks. And grouse, thanks plus for that last link.
posted by foxy_hedgehog at 9:22 AM on August 8, 2007

No one is trying to limit access to the scientific community, they're just trying to make money. LD50 information should be in Material Data Safety Sheets, and you can find many MSDSs on the web, but many of the sites require registration.

Manufacturers are required by law to supply the purchaser with an MSDS. There are literally millions available on the web for free, just make sure you are looking at the same concentration.
posted by Big_B at 9:22 AM on August 8, 2007

Just checking in to see if there are any matters not covered.

One is that LD50 is a poor measurement of toxicity. There are so many other things to take into account - route of exposure, length of time of exposure, length of time to dying. You can have a substance that will kill 50% of subjects but take years to do it, cancer-inducing for example. Most studies in animals are acute exposure.

Another problem is that human drug reponses don't necessarily fall on a Gaussian curve. The dose that will kill those with a common genetic mutation to handling succinylcholine will be very different from those that don't have that trait.

Another problem with LD50 is that it is talking about death only - not severe or crippling toxicities.

Don't be fooled by LD50. It is way too little information to know a compound's toxicity.
posted by dances_with_sneetches at 9:35 AM on August 8, 2007 [1 favorite has favorites]

But it does tell you that Dimethoate (215) might be a better choice than Guthion (6) to spray on your apples, although I would use neither.
posted by weapons-grade pandemonium at 10:10 AM on August 8, 2007

Dances, in re:

One is that LD50 is a poor measurement of toxicity. There are so many other things to take into account - route of exposure, length of time of exposure, length of time to dying. You can have a substance that will kill 50% of subjects but take years to do it, cancer-inducing for example. Most studies in animals are acute exposure.

Another problem is that human drug reponses don't necessarily fall on a Gaussian curve. The dose that will kill those with a common genetic mutation to handling succinylcholine will be very different from those that don't have that trait.

Another problem with LD50 is that it is talking about death only - not severe or crippling toxicities.

Don't be fooled by LD50. It is way too little information to know a compound's toxicity.

What sort of standard datum is a better way of measuring such lethal toxicities? How would a laylady like myself find it?
posted by foxy_hedgehog at 10:34 AM on August 8, 2007

The dose that will kill those with a common genetic mutation to handling succinylcholine will be very different from those that don't have that trait.

Dances answer gives a very good example of why "toxicity" is not an all-or-nothing phenomenon that can be boiled down to a single number; you have to understand what a compound does to the body in order to determine how lethal it is. For those who don't know what it is, succinylcholine is a drug that is used to briefly paralyze patients to allow for easier endotracheal intubation, as during anesthesia. As long as the patient's breathing is supported (the diaphragm and other respiratory muscles are paralyzed after a dose of the drug) there is actually little additional toxic effect in humans and even a massive overdose will not be fatal. As mentioned, there are exceptions. Sux (as it is called in medical jargon) is a trigger for malignant hyperthermia which is almost always fatal if untreated and is the result of any of a number of related mutations affecting something known as the ryanodine receptor. A similar, potentially fatal reaction can occur in patients with certain types of muscle disease and is the subject of a "black box" warning by the FDA. Finally, there are people who have one of a number of mutations in the gene for cholinesterase, the enzyme that breaks down sux and many other drugs. These people will not have a life-threatening reaction to the drug, but will remain paralyzed for much longer than usual following a dose and will require artifical ventilation during this time. They will recover completely, however. As you can see, what the "lethal" dose of succinylcholine is will vary from person to person and in most people there is no known lethal dose as long as their breathing is supported. So the answer to your question really depends on what drug or compound you are asking about and what it does to the body. A very different example would be a drug like digoxin that has direct toxic effects on the heart at doses not much higher than those given therapeutically, and so has a much better defined LD 50 in humans (and overdosages are common enough that there is a specific antidote).
posted by TedW at 11:24 AM on August 8, 2007

Ted, thanks for that- I've been put out with sux before, which makes this all the more intriguing.

So this technical detail is really just speculative guesswork, dancing around an idea within the constraints posed by experimentation.
posted by foxy_hedgehog at 8:13 PM on August 8, 2007