Does mild exercise cause ALS? (Is this study valid?)

Triathlon Forum
1

There is a study which will be making the rounds of media outlets over the coming weeks due to is claims that exercising as little as 15 min per day 2 times a week causes ALS: Motor neurone disease: Intense exercise increases risk, say scientists - BBC News. I am already being bombarded by my mother-in-law about the horrors of training and I will guess others will be asked the same questions. My question is whether the study is fundamentally flawed because of its methodology or if epidemiologically it hold water. I am not dismissing the theoretical mode of action or the potential for this to occur but I have serious questions over whether the study provides any evidence one way or another.

The study used the following methodology:

  1. They screened the results of a large(ish) gene study for ‘exercise-associated transcriptome changes.’
  2. They identified the prevalence of ‘ALS risk genes’ within the above data set.
  3. They used the results of a questionnaire to split the data between people who had exercised at least 15min 2x a week over the past four weeks and those who hadn’t. (12577 people in the exercise group and 23475 in the non-exercise group)
  4. They statistically compared the ALS risk gene prevalence between the two groups.

Based on the above I have the following questions:

  1. Did their initial screening phase fundamentally bias the results? By looking only at only what they believed to be exercise-associated changes rather than looking at total ALS gene prevalence did they not bias the results before they began? Also wouldn’t the correct way to report the data to be to normalize every individual for the # of exercise associated changes rather than bin the data base on an arbitrary exercise cut-off?

  2. How do they know the total exercise loads of their two populations were significantly different? Its a well established in the medical community that when asked by a doctor people don’t provide accurate or precise information on how much they weight, how much alcohol they drink and how much they exercise. You have to take all the answers as a ballpark figure or validate them and by setting the bar so low its not apparent to me they achieved a significant split between the groups. I would have though you needed to go much higher on exercise load to be confident you are differentiating people who do and don’t exercise. Even if we assume people were 100% accurate and precise by only looking at a 4-week window you are getting a snap shot and I have no idea how representative a 4-week window is of life-long practices.

3). Does risk gene prevalence correspond to higher risk? This may be a dumb question but I have always believe ‘potency’ comes into play because that not all gene copies are equal in effect. For example with breast cancer I was under the impression that a lot of genes are involved but huge emphasis is placed in the brca gene because certain varieties of that gene can dominate the total risk. So wouldn’t you need some sort of weighted risk indicator that takes into account potency?

There does appear to be enough circumstantial evidence around ALS and exercise to hypothesis about a link and do more research. However I think bad research (or poorly thought out research) does more harm than good by producing misleading data. I am not sure where to put this study.

2

I also saw this, this morning.

As is (unfortunately) often the case, two things were in my mind

  • research that end ups with vague un-usable information (I.e. WTF are we supposed to do then with this new knowledge ?) and becomes the selling pitch for… needing more research funding to go away and do some more research (who’d have expected researchers wanting funding for more research?)

  • shiiite scare-headline low quality journalism reporting on it. (These are the same lazy feckers who in the UK propagated the fear about MMR vacines causing autism which put the benefits of vaccination back 2 decades… and led to the rise of M, M and R in children- some severe).

Only buried in the text is the ‘but on balance keep exercising’) as we all know instead we’ll have a nation of vastly overweight lazy arses who instead die of heart disease, diabetes or a stroke from high blood pressure.

Still, I suppose it’s a distraction from the usual poor (mis)information, scaremongering and lack or critical analysis in the same media about Covid (to be clear, I’mnot a Covid denier. Far from it. But I despair at the lack of analysis of manipulated ‘data’ or wider view of risk management / balance of risk by the media).

3

The study used the following methodology:

  1. They screened the results of a large(ish) gene study for ‘exercise-associated transcriptome changes.’
  2. They identified the prevalence of ‘ALS risk genes’ within the above data set.
  3. They used the results of a questionnaire to split the data between people who had exercised at least 15min 2x a week over the past four weeks and those who hadn’t. (12577 people in the exercise group and 23475 in the non-exercise group)
  4. They statistically compared the ALS risk gene prevalence between the two groups.

So many times looking at some risk factor or an underlying factor that is suppose to lead to disease doesn’t.

Maybe it’s true that exercise is associated with ALS, if that is the case why not look at people who get ALS and see if they are more likely to be regular exercisers or engage in more physical activity than those that don’t. Futzing around with ALS risk genes seems like a good way to draw some incorrect conclusions.

I’d also think if the effect size was large, neurologists would have noticed a long time ago that people who get ALS tended to be runners, etc.?

4

did anyone look at the prevalence statistics for ALS? In western countries it is about 2/100,000 and in US it is about 5/100,000 source: https://www.cdc.gov/mmwr/volumes/67/wr/mm6746a1.htm
for context 1 in 4 of all causes of death in the US are heart related issues also sourced from CDC (About **655,000 Americans **die from heart disease each year—that’s 1 in every 4 deaths.). So I would worry way more about inactivity re heart disease than exercise related ALS…

just poor quality reporting and probably poor scientific rigour.

5

did anyone look at the prevalence statistics for ALS? In western countries it is about 2/100,000 and in US it is about 5/100,000 source: https://www.cdc.gov/...s/67/wr/mm6746a1.htm
for context 1 in 4 of all causes of death in the US are heart related issues also sourced from CDC (About **655,000 Americans **die from heart disease each year—that’s 1 in every 4 deaths.). So I would worry way more about inactivity re heart disease than exercise related ALS…

just poor quality reporting and probably poor scientific rigour.

You’re about twice as likely, even if fit, more so if not fie, to have a cardiac event during exercise than at other times. So none of use should be exercising!

But of course, the protective effects far outweigh this acute increase in risk since regular exercisers die from cardiovascular disease at much lower rates than non exercisers.

6

just poor quality reporting and probably poor scientific rigour.

The article linked in the OP seems pretty reasonable to me. I’m not qualified to judge the science itself. The article does stress multiple times that no one should stop exercising because of this, including the second sentence.

7

The issue with the approach you suggested is that ALS is fatal and in order to die of ALS you must survive all the other potential causes of death.

ALS typically hits people between 50-70 and exercise is a associated with a massive reduction in the main causes of death in this age group. So basically if you exercise you are statistically more like to be killed by ALS because you survived the other things that would have killed you. That does not mean exercise causes ALS it just means exercise helped you survive long enough to develop ALS. You need someway to identify people who would develop ALS at 70 but die of heart failure at 55 due to inactivity and therefore are spared from ALS.

I am sure smarter people than me have figured out a way to do this. I remain unconvinced the authors of the paper are those smarter people.

8

it is a sentence from the BBC article that gives me pause to question their care in editing and thinking this over:" They used a technique called Mendelian randomisation to turn that data into an experiment, and showed people whose DNA makes them more likely to do strenuous activity were more likely to get MND." in particular the part:“showed people whose DNA makes them more likely to do strenuous activity” I was unaware that my DNA made me do strenuous exercise! I sort of stopped reading at that point.

The research paper seems to say that a specific subset of ALS patients (those with a specific genotype) would be more at risk if they did strenuous and frequent exercise. I think even the paper title might be somewhat misleading, the risk increases for a subset of those who are genetically predisposed to ALS.

The paper’s conclusion is:
In conclusion, the current evidence supports a complex causal relationship between physical exercise and ALS. However, it is clear that, for the majority of individuals, the health benefits of a physically active lifestyle markedly outweigh the risks. The key objective for future research is to understand which individuals are at risk of developing ALS if they exercise excessively and provide appropriate lifestyle counselling. Our work goes some way towards developing this aim and in particular, we propose that C9ORF72 penetrance maybe influenced by high levels of physical activity.

9

it is a sentence from the BBC article that gives me pause to question their care in editing and thinking this over:" They used a technique called Mendelian randomisation to turn that data into an experiment, and showed people whose DNA makes them more likely to do strenuous activity were more likely to get MND." in particular the part:“showed people whose DNA makes them more likely to do strenuous activity” I was unaware that my DNA made me do strenuous exercise! I sort of stopped reading at that point.

As a former Anthropology professor of mine used to say, “the gene that makes people research genes strikes again!”

10

it is a sentence from the BBC article that gives me pause to question their care in editing and thinking this over:" They used a technique called Mendelian randomisation to turn that data into an experiment, and showed people whose DNA makes them more likely to do strenuous activity were more likely to get MND." in particular the part:“showed people whose DNA makes them more likely to do strenuous activity” I was unaware that my DNA made me do strenuous exercise! I sort of stopped reading at that point.

The research paper seems to say that a specific subset of ALS patients (those with a specific genotype) would be more at risk if they did strenuous and frequent exercise. I think even the paper title might be somewhat misleading, the risk increases for a subset of those who are genetically predisposed to ALS.

The paper’s conclusion is:
In conclusion, the current evidence supports a complex causal relationship between physical exercise and ALS. However, it is clear that, for the majority of individuals, the health benefits of a physically active lifestyle markedly outweigh the risks. The key objective for future research is to understand which individuals are at risk of developing ALS if they exercise excessively and provide appropriate lifestyle counselling. Our work goes some way towards developing this aim and in particular, we propose that C9ORF72 penetrance maybe influenced by high levels of physical activity.

I didn’t bother to read the article, but the fact that they use the term “causal relationship” in the context of an observational study should raise massive red flags that indicate B.S. Correlation is not causation and I don’t see any natural experiment that would provide some assurance that this isn’t complete nonsense.

11

it is a sentence from the BBC article that gives me pause to question their care in editing and thinking this over:" They used a technique called Mendelian randomisation to turn that data into an experiment, and showed people whose DNA makes them more likely to do strenuous activity were more likely to get MND." in particular the part:“showed people whose DNA makes them more likely to do strenuous activity” I was unaware that my DNA made me do strenuous exercise! I sort of stopped reading at that point.

The research paper seems to say that a specific subset of ALS patients (those with a specific genotype) would be more at risk if they did strenuous and frequent exercise. I think even the paper title might be somewhat misleading, the risk increases for a subset of those who are genetically predisposed to ALS.

The paper’s conclusion is:
In conclusion, the current evidence supports a complex causal relationship between physical exercise and ALS. However, it is clear that, for the majority of individuals, the health benefits of a physically active lifestyle markedly outweigh the risks. The key objective for future research is to understand which individuals are at risk of developing ALS if they exercise excessively and provide appropriate lifestyle counselling. Our work goes some way towards developing this aim and in particular, we propose that C9ORF72 penetrance maybe influenced by high levels of physical activity.

I didn’t bother to read the article, but the fact that they use the term “causal relationship” in the context of an observational study should raise massive red flags that indicate B.S. Correlation is not causation and I don’t see any natural experiment that would provide some assurance that this isn’t complete nonsense.
and of course don’t forget it needs more study or as they say future research… meaning they are writing the grant proposal as we speak! I think this is a csae of we did the study so we had to come to some conclusion. so if it is 7 in 100,000 of the population that might end up with ALS then a fraction of that 7 we are heading towards 1 in 1,000,0000 real fast, so if it were the US that would mean 384 individuals maybe have the genotype? I am sure it matters to them that if they could avoid strenuous exercise and improve their life that sounds like a bet worth taking.

12

The article is available here: https://www.sciencedirect.com/science/article/pii/S2352396421001900 about the link between exercise and motor neurone disease/ ALS.

I’m not a scientist but do like to try to understand stuff like this. Motor neurone disease is horrible and so anything to help understand how to avoid it sounds good.

Like the authors say it’s known that for the vast majority of people, the health benefits of a physically active lifestyle greatly outweigh away risks.

But on the other hand, the possibility of a link between exercise and motor neurone diease has been talked about for a while. One of the questions is whether exercise is actually the cause, or just an association.

What these guys did was to take data from the UK biobank project - a large database containing genetic and health information from about half a million people- and use a technique called Two-sample Mendelian Randomisation (MR) to try and answer this question.

This is where I run up against the limits of what I understand, but there’s a good explanation of what MR involves here https://www.bmj.com/content/362/bmj.k601

MR gets round things like recall bias in studies, by using genetic variation at birth as a “natural experiment” to investigate the relationship between a modifiable risk factor, and a health outcome. It allows researchers to use observational data to help answer the question “is this behaviour/risk factor likely to be causal for this disease/outcome?”

In this motor neurone disease study, the researchers used this approach.

They took the data from the UK biobank project, and looked for a number of genetic variations (single nucleotide polymorphisms) which occurred more commonly in people who declared that they had exercised strenuously 3 or more times per week in the last 4 weeks.

Then they looked at genetic variations in a different sample of people who have motor neurone disease.

Then they use software to bring together the associations found in these 2 sample groups, and somehow conclude there is a degree of probability that exercise is causal for motor neurone disease.

Now to a statistician that may be a valid thing to do, but I think it requires a degree of faith in every step of the process. I have read the text of the study twice, and do not really understand how this conclusion has been reached.

My main hesitation is around the composition of the first sample group. Exercising for 20 minutes, three times per week, for the last 4 weeks feels like quite a low bar. And then, the scatter graphs in the study suggest that a large number of genetic variants (single nucleotide polymorphism) are weakly associated with increased exercise behaviour. I just don’t know what to make about this. It seems like a stretch to say that if these same variants are linked with motor neurone disease, then it is the exercise which causes the disease.

Does anyone else here see this differently? Please explain if so.

Later the article refers to a small case-control study which investigates the gene-environment link further. A small number of people with motor neurone disease were asked about their exercise history. For those with a certain form of motor neurone disease who also had a certain genetic picture (a C9ORFJ2 expansion), the age at onset might be linked to lifetime exercise. This part I have less difficulty accepting, but sgain, this does not say to me that exercise causes MND, just that it may have accelerated the condition for people with one form of the disease who also had a particular genetic picture.

Like others have said I think the media ought to be careful and balanced in how they report this. Science advances by small steps & the jury is stil out here - some other studies have not suggested a causal association.

But we do know that exercise:

  • Helps you control your weight
  • Reduces your risk of heart disease
  • Reduces the risk of type 2 diabetes
  • Improves mental health and mood
  • Help keep your thinking, learning, and judgment skills sharp as you age.

And so we want to encourage people to do it.

13

and of course don’t forget it needs more study or as they say future research… meaning they are writing the grant proposal as we speak!

Studying the complex factors causal and correlative factors (genetic, environmental, and behaviorial) related to ALS is a noble undertaking. I really hope that good people apply for funds for future research. I trust the scientific community in general to vet proposals, and eventually snuff out any weak science. I’m in general optimistic about this “big data” style of observational research for some types of study that just can’t be done any other way. Though, as I said, can’t comment on the quality of this particular study.

14
  1. They screened the results of a large(ish) gene study for ‘exercise-associated transcriptome changes.’

I had a quick look at their data set and biased experimental hypothesis (biased loci selection), that if one could easily pick a different set of loci (negatively associated with ALS) and come up with the exact opposite hypothesis.

When one can show that their loci selection translates to proteins that are actually promoting neuroinflammation (by proteomics and actual biology ), then it may be worth to revisit their data and approach.

.

15

Why are we being like this?

Literally second line in the linked article is:

The team at the University of Sheffield said nobody should stop exercising as a result of their study.

My emphasis. So we don’t need to argue that exercise isn’t bad. No-one is saying otherwise.

in particular the part:“showed people whose DNA makes them more likely to do strenuous activity” I was unaware that my DNA made me do strenuous exercise! I sort of stopped reading at that point.

If I read that particular DNA made some people more likely to be lazy I wouldn’t blink. So not a huge leap to more likely to do strenuous exercise. Not making you do it, making you more likely to do it. But if you want to disagree with that point then see the referenced study and pick it apart. This is contradiction not an argument.

Surely this is the way science works? Someone highlights a finding and following research takes it further or challenges it both of which are progress.

Since no-one is saying don’t exercise and one of the possible benefits is to find a way of screening for people at risk I don’t see what has got people worked up.

16

Oddslug, I have looked at the data in the referenced study.

This is some of the data. On the horizontal axis, some single nucleotide polymorphisms which were selected because they appear more often in exercisers. On the vertical axis, whether there appears to be any negative or positive association between those SNPs and motor neurone disease.

I think this shows that some of the genetic markers that are slightly more common in people who exercise, are positively associated with MND. And some are negatively associated.

Do you see a strong correlation there? I am not sure I do at first glance.

Looking at the statistical analysis, they explain that several approaches (instruments) were used to see if there is correlation. Only one of these suggests a significant relationship at p=0.01. This was described as a “liberal instrument using an inverse variance method of Mendelian Randomisation.”

Several other approaches (instruments) appear not to have shown a significant relationship. If I am understanding right.

I agree that this adds something to the pool of scientific knowledge, and that is good because it is progress. But am sad to see an “exercise causes motor neurone disease” message appearing in the mainstream media. That feels like a very blunt interpretation of this science.

Screenshot_2021-06-13-07-12-40-402_com.google.android.apps.docs.jpg

17

Nothing to add here, but wanted to say that I am enjoying the calibre of this discussion
.

18

just poor quality reporting and probably poor scientific rigour.

The article linked in the OP seems pretty reasonable to me. I’m not qualified to judge the science itself. The article does stress multiple times that no one should stop exercising because of this, including the second sentence.

  1. As Richard Feynman would say, “one study does not prove a guess (hypothesis) is right, only that it’s not wrong”. But further studies could prove it wrong.
  2. Correlation is not causation. To suggest a causal relationship with this study is not scientific. At best, correlation can suggest a hypothesis, but not prove the hypothesis.
  3. The source of the exercise data is a questionnaire. Ha. Good luck with that. Garbage in, Garbage out.

There is a lot of bad science being done. Much of it is observational epidemiology. If I had a dollar for every time I saw two different observational epidemiology studies with opposite outcomes, I’d be a wealthy man.

Look beyond the obvious. Use critical thinking. Watch Feynman’s video on YouTube on the scientific method. “Science is the belief in the ignorance of experts.” Put another way, “no opinion is better than expert opinion when no scientific data is available.”

Cheers

19

There is a lot of bad science being done

Sure. But is his paper bad science? I’ve seen claims of issues with causation vs. correlation. But no evidence that anyone’s actually demonstrated it beyond speculation. E.g. the criticism here is interesting, but not at the level of an actual peer review.

I understand skepticism of the the questionnaire. I’d agree it’d be preferable to outfit all the study participants with activity trackers for a year or so, etc. But I understand the resource issues with doing that. And I don’t think it’s totally unreasonable for people to self-report broad activity levels in a way that gains reasonable statistical validity given proper treatment.

And regarding observational studies, I understand that as well. But for some topics of study there is simply no other type of study you can do. Hence my optimism for the use of “big data” statistical techniques in observational study. For hard problems that have remained intractable to other forms of inspection. ALS might just be such a thing. E.g. there’s no simple set of genes that you flip on, and you get ALS (except for the relatively small % who have a definitively inheritable version). Or no single chemical you get exposed to, and you get it. It might have really complex causes that you can only unravel through observation that covers a wide range of behaviors, environments, and genetic makeups. I think (hope) that we do lots more of these types of observational studies to try to gain insight into human health. I agree we have to be really, really careful in treatment of the strength and meaning of the outcomes. And I’m perfectly willing to throw this particularly study under the bus if it’s really demonstrated to have been poorly conceived or conducted.

20

All of the real scientists I’ve talked to (I am not one), will tell you correlation is not causation. I believe that is well understood, but I suppose some would argue with that. However, lack of correlation can prove there is not causation.

I understand that doing interventional studies is hard and costly. But just because they are hard doesn’t make observational studies robust. Their limitations must be understood, and thus their conclusions subject to greater scrutiny. Furthemore, these can be gamed. Think about this, if you received a questionnaire and were asked what you ate over the last year, how accurate do you think it would be? I was actually listening to a podcast today where they sighted data on how inaccurate those can be. One example, people significantly underestimate calories consumed, and those are often processed food calories. Also, just because it is peer reviewed doesn’t make it a good study.

Thankfully, this is about to change. With activity monitors and continuous glucose monitors (CGM), we will be less reliant on survey data, and will get measurable data. This is where big data analytics will be useful. Instead of unreliable questionnaire data, with limited sample size, we will have orders of magnitude more data, that is of much higher quality. When somebody says they don’t eat sugary foods, it will be easy to say, then why does your glucose level spike every night at 8:00pm? I suspect this will be a revolution. I also suspect that many of the things we’ve been told, will be proven to be false. There is a lot of money involved.

When I can buy a smartwatch with a built in CGM, that is a game changer. Note: I’m gonna to do a one month test with a CGM. I’ll pay the out of pocket cost for that time period.

Cheers.