Saturday, May 18, 2013

No Data on Environmental Causes of Type-1

I'm not completely happy with this posting, but I've been working on it for many weeks, and at this point I think it is better to publish what I have, than to continue to struggle with it.  Maybe in the future, will post an improved version.

This posting is a little different from my usual fare.  Instead of discussing specific research results, I'm going to discuss, in more general terms, a subject that has come up repeatedly:

What is causing the increase of type-1 diabetes?

Every six months or so we get a study, government report, or newspaper article claiming that there must be something in the environment which is causing the number of type-1 diabetics to go up, way up. These articles usually follow the same path:
  1. Description of type-1 diabetes as having a genetic and environmental cause [d0].
  2. The number of type-1 diabetics is going up (way up!)
  3. The number of younger type-1 diabetics is going up even more than older ones.
  4. So there must be an environmental cause (usually a toxic chemical) which is growing, and causing the numbers of type-1 diabetics to grow.
They often contain a quote like this one:
Despite a strong genetic component to the susceptibility of T1DM, this marked increase in incidence in different populations within a short period of time cannot be explained by increased transmission of T1DM susceptibility genes. So what is the cause? [r1]
Or like this one:
Essentially all researchers agree that changes of this magnitude cannot be explained by genetics alone. [r6]
(The d-numbered footnotes refer to extra discussion and r-numbered footnotes refer to references, both are at the bottom of the posting.)  Before you get too worried about such news reports, it's best to think about what should be happening.  The simple answer is: of course the rates of type-1 diabetes are going up, and we should all expect that.  Why?  Type-1 diabetes has a large genetic component.  Before 1920, most people who were diagnosed with type-1 diabetes died before they had children.   Now, those people are living longer, happier, richer lives, with children and grandchildren of their own.   So the genetic component of type-1 is becoming more common, and of course the rate of type-1 diabetes is going up [d1].  We certainly would not expect it to be going down or staying the same.

So the the correct response when someone says the rate of type-1 diabetes is going up (way up!) is to yawn, and say "Of course, I would not expect anything else" and "Isn't it great that what was previously a fatal disease is not any more".   However there are still some issues which need some more discussion:
  1. Referring to age at diagnosis: why is the rate of very young type-1 diabetics going up more than the rate of older diabetics?
  2. Is there something in the environment which is contributing to the increase of people with type-1 diabetes, and especially a toxic chemical of some kind?  Even if some of the growth rate is caused by genetics, maybe some of it is not?  
  3. Are sentences like "the increase is too high to be explained by genetics alone" (or similar) supported by the data we have, or are they fear-mongering?
Why are more children getting diagnosed at a younger age?

I'm not going to review it in detail, but I think there is pretty strong evidence that more children are getting diagnosed with type-1 diabetes at a younger age than previously.  For example [r1,r2] both show this. The percentage diagnosed between ages 0 and 5 is going up faster than the more traditional diagnosis between 10 and 20 years old.

The first thing to realize here, is that you would expect this behavior based on the genetics and impact of type-1 diabetes.  Type-1 is not (usually) caused by a single gene.  It is caused by complex groups of genes which, acting together, make it more likely that you will get type-1, and other groups of genes that make it less likely. It is likely [need to specific references here] that there are gene groups that tend to cause type-1 in younger kids, and other gene groups that tend to cause type-1 in older kids.  Prior to the 1920s, everyone who had type-1 diabetes died very close to diagnosis.  So if you had the genes for the age 0-5 type, you had zero chance of having children.  But if you had the genes for a 10-20 year old diagnosis, then maybe you would have children (especially in earlier times, with younger parents).  Obviously, this would be a very small number of births, but it would be higher than zero.  So the genetic "filtering" that type-1 death caused, was stronger for those "younger" genetic variants.  Therefore when the genetic "filtering" is removed, we should expect type-1 diabetes to rise more quickly in younger kids than older kids.  And that is exactly what is seen.

Summary: faster growth in type-1 diabetes diagnosis in younger kids is something that we should be expecting based on the known genetics and lethality of the disease.

Is there something in the environment causing more type-1 diabetes?  
Maybe a toxic chemical?

Obviously, there are a huge number of chemicals out there, and any one of them might increase the incidence of type-1 diabetes.  Even if we proved 1000 (or 10,000) chemicals do not cause type-1 diabetes, it might always be caused by one we have not tested.  So the possibility of chemicals causing type-1 diabetes will always be with us.  But I think there are two important points that should be made here:

First, it is easy to show that a chemical causes type-1 diabetes.

We have an animal model of type-1 diabetes.   It's called the NOD mouse, and has been in widespread use for decades [d3].  The number of these mice who will naturally come down with type-1 diabetes, and when it happens, is well known.  And some of these mice won't come down with the disease, even if most do.  Therefore, any chemical you want to test, you can just give it to these mice, and see if the rate of type-1 diabetes goes up, or if the mice get it more quickly, than the untreated mice. [d4]

This is the kind of mouse-based research that is done all the time, and it isn't even that expensive.  Running a test on 300 mice is vastly cheaper and faster than running a test on even 1 person.

So if anyone really believed that they know which chemical or drug caused type-1 diabetes. They could become famous quickly, easily, and inexpensively by testing their theory in mice. The fact that no one has done this suggests to me that none of the common chemicals sometimes suggested as a cause of type-1 diabetes, really do cause it [d5].

Now, some people will make excuses.  They are likely to say "we can't test chemical X because no one will fund it", but that is untrue for a number of reasons.  Most obviously: non-profits, environmental groups, and government agencies often fund safety research.  But also, companies that produce competitors to a chemical have a strong economic motivation to fund research that the other chemical is unsafe, since that would lead to higher sales of their (competing) chemical.  Finally, remember that fame is an important motivator (in addition to money), and providing strong evidence that a common chemical is unsafe is a sure path to fame in the university, non-profit, and government worlds.

Here are examples of experiments of that kind (could show chemical danger in NOD mice) but did not find any problems:

Trichloroethylene (TCE): [d6]
Mercury:  [d7]
Bisphenol A (BPA) is a more complex case, which I may cover in a future posting (time permitting).

Second, not a lot of chemicals fit the dosing/timeline profile.

The second problem with blaming a toxic chemical, is that the chemical needs to have a use profile that matches type-1's growth profile, in those countries where we have good data on type-1's growth profile.

For example, it couldn't be DDT.  Because in the USA, DDT use skyrocketed right after World War II, and then dropped to nearly zero in the 1970s.  But the rate of type-1 diabetes continues to grow at a pretty constant rate.  There was no big increase right after WWII and no dropping off in the 1970s.  Similar arguments can be made against lead, PCBs, BPAs, and many other toxins [d9].

But it turns out that type-1 diabetes growth appears to be pretty constant.  More kids are diagnosed each year, but the increase is linear, and there are no big troughs or hills.   (Although [r6] shows some small bumps up and down.)  Indeed, if you look at the whole time period from 1920 to 2012, the growth is pretty constant the whole time (as much as we have data for). [Need to add specific references for this.]  I think that suggests that there is no chemical environmental cause, because the environmental chemicals common to the 1920s are not common now, and visa-versa.  I just don't see a good candidate.  However, since testing the chemical is cheap and easy (see previous discussion), the moment someone does identify a chemical who's use has been steadily growing for the last 90 years, without any large troughs or hills, that chemical could be quickly tested.  And if it happened to match the small ups and downs in the [r6] data in the same countries where that data was gathered, that would be even stronger evidence.

Based on all this, if there is an environmental factor, a toxic chemical is not my leading contender.  I think something like the "hygiene hypothesis" is more likely, or maybe something in the diet, or something related to affluence [d8].

The increase is too high to be explained by genetics alone.

Think about this claim for a minute.  This sentence is based on three pieces of data:
  1. We know how much type-1 diagnosis is increasing.
  2. We know how much it should be increasing, based on genetics.
  3. Therefore, we can see that there is a gap, which must be filled with an environmental cause of growth.  
Of these three statements, only the first is true.  See [r1,r2, r6] and other studies.  Even there, our knowledge is imperfect, but we do have some data about the overall change in the rate of type-1 diagnosis.  But for the next two, we have nothing; absolutely nothing.  Not all the genes that lead to type-1 diabetes have been identified.   Not all the genes that protect against type-1 have been identified.  How these genes interact to cause or avoid type-1 diabetes is not known.  In short, we have no idea, what the levels of type-1 diabetes "should" be, based on genetic susceptibilities.  Obviously, since we don't know 2, we can't know 3.  In my opinion, people who say "there must be an environmental cause for the increase in type-1" are fear mongering, or looking for more research funding. 

Remember: I agree that there is an environmental cause of type-1 diabetes, but I'm also saying that there is no evidence that it is growing. Put another way (with double negatives): I'm not saying that there is no environmental component that causes type-1 diabetes. We know there is [d0]. Rather, I'm saying that there is no evidence that this environmental component is causing the increase in type-1 cases. It is not growing or causing more cases of type-1 diabetes.

Summary:  There is evidence that an environmental component exists, but there is no evidence that the environmental component is causing growth in type-1 diagnosis.

Looking at Genes Specifically

As far as I know, no researcher is looking at the overall genetics of type-1 diabetes to see if the overall genetic change is causing the overall type-1 diagnosis change.  However, there are a couple of studies that look at a single gene specifically [r7].  These studies both looked at a gene highly associated with type-1 diabetes [d10] and they both found that specific gene had become less common in the population.  Not more common, as would be expected.  One study looked at the period from the 1960s to the 2000s, and the other compared the 1980s to the 2000s (roughly).

Proponents of an environmental cause to increases in type-1 diagnosis can point to these studies as support for the idea that the growth is not caused by genetics.  However, these studies provide only the tiniest support.  Neither of these studies actually looked at the total genetic change affecting type-1 diabetes.  Indeed, they couldn't do that, because we don't yet know all the genes that affect type-1 diabetes.  It's like trying to guess if the value of pocket change is going up or down, based on the number of nickels in people's pockets, when you don't even know all the kinds of coins in circulation.

So while these two studies are scientifically interesting, and should form the basis of more research, they are nowhere near the level of knowledge we need to say that the current growth in type-1 diabetes diagnosis is caused by an environmental factor.


[d0] We know that the cause of type-1 diabetes has a genetic component, because it is more common in people with relatives who also have type-1 diabetes.   In particular, the closer a relative who has type-1 diabetes, the more likely a person is to have it.  This is a classic sign of a genetic component.  However, we know that there is an environmental component as well, because of "twin studies".  For a purely genetic condition, there should be no cases of identical twins, one with type-1 diabetes and one without.  However, there are.  In fact, in less than half of identical twin pairs do both twins have type-1 diabetes.  It is more common that only one does [r3].  This difference in disease between identical twins is a classic sign of an environmental component.  So it is clear that type-1 diabetes is caused by a combination of both genetics and environmental causes.

[d1] No matter how I phrase it, some people read those last two sentences, and think I'm "blaming" parents for their children's type-1 diabetes.  I'm not, and that's a silly idea in any case.  I'm discussing genetics.  It has nothing to do with blame.  No one decides what DNA to pass to their children.  And it is completely unreasonable to think "well something might go wrong with my kid, so I won't have any."  If people thought like that, no one would ever have children, because something can always go wrong.

[d3] Technically, we have at least three animals models for autoimmunity-based diabetes.  Another is the BB ("Biobreeding") rat [r4], and the third is an animal that given a small dose of a beta cell toxin.  The beta cell die off somehow triggers autoimmunity.  This last technique is quite different that given the animal a large dose of a beta cell toxin (which is more common).  Giving a large dose kills the beta cells, but does not trigger autoimmunity.  It can be used to test insulin replacement, but not autoimmune effects.

[d4] I know a lot of people are frustrated with NOD mice, because cures that work in those mice have failed when used on people.  However, in this case, we are using NOD mice to show danger, rather than improvement.  The "standard of proof" is much lower to show danger, as compared to showing safety and effectiveness.  To show safety and effectiveness most people would require several tests in animals followed by several tests in people.  On the other hand, to show danger, even one or two tests in animals would be enough to convince most people that a chemical was unsafe.  The big frustration of NOD mice (that cures don't work when tried in people) doesn't affect this kind of danger testing, because if the chemical is found unsafe in mice, it's done: no tests in people are needed.

[d5] Basically, the first time I hear someone say that chemical X might cause type-1 diabetes, I look at a calendar.  I then wait a few years.  If no one has published results for the obvious test in NOD mice, then I'm pretty sure chemical X does not cause type-1 diabetes.  I then wait a few more years.  By that point, if no one has published results, then I think it's settled that chemical X does not cause type-1.  A logician will continue to chant "absence of evidence is not evidence of absence," but in the real world, it is.

[d6] A quote from their abstract:
To test whether TCE [Trichloroethylene, a chlorinated hydrocarbon] can exert similar deleterious effects on organ-specific autoimmune diseases, non obese diabetic (NOD) mice were given 5 mg/ml TCE via the drinking water for 12 weeks. ... Contrary to what has been found in systemic models of autoimmunity, TCE did not accelerate the diabetes of NOD mice and may have a protective effect.
[d7] A quote from their abstract:
We found that three weeks of treatment with mercury was also able to significantly suppress the development of insulitis and postpone the onset of diabetes in these mice. Thus, mercury-induced immune activation can counter-regulate the Th1 cell-mediated autoimmune responses and confer a partial protection against autoimmune diabetes in NOD mice.
(Although I don't think people will line up for a clinical study injecting mercury or TCE to delay onset of type-1 diabetes. :-)

[d8] I am not saying that I think the "hygiene hypothesis" is correct. I am saying something much weaker: that the evidence we have for the "hygiene hypothesis" is stronger than the evidence we have for any single, specific chemical causing type-1 diabetes.

[d9] This line of reasoning also tends to exclude breast feeding as an environmental protective against type-1 diabetes.  Breast feeding rates were high in the 1920s, dropped to their lowest in the 1960s, rose in the 1970-1990s, and have been drifting higher since then [r5].  None of that is reflected in type-1 diabetes rates.  If breast feeding protected against type-1, then the type-1 rate would be dropping slightly right now, not rising.  Also, it would have been much higher in the 1960s, than in the 1980s, but the reverse is actually seen.

[d10] I'm not a geneticist.  Both studies dealt with something called "HLA class II".  One study looked at a gene called "HLA-DR, DQ" , while the other looked at a genotype called "HLA-DR3/4-DQB1*0302".  I'm not sure if they were both looking at the same gene, or slightly different genes.  But in either case, they are clearly only looking at a tiny part of the genetics that lead to type-1 diabetes, which is my main point.


[r1] Incidence of childhood type 1 diabetes: a worrying trend by Ronald C. W. Ma and Juliana C. N. Chan
October 2009

[r2] Type 1 Diabetes in Urban Children Skyrockets, Increasing by 70%
Jan 22, 2013

This Italian twins study found:
Genetic contribution to type 1 diabetes susceptibility was 40%, and the shared and individual-specific environmental components were 51% and 9%, respectively.  [I removed the confidance intervals from this sentence, but they are in the original abstract if you want them.]

[r4] Wikipedia on BB rat:

From 200-2008:
From 1850-2000: (see Figure 3)