I think the article says iron levels are negatively correlated to longevity, but "the association might be due to a already-known correlation to haemochromatosis" which I have no idea what that is. I presume it's some known population that's old that happens to also have an enrichment of haemochromatosis but where individuals in that population without it do not show a positive defect in longevity?
Due to them saying that "the association might be due to an already-known correlation to haemochromatosis" I'm assuming they didn't have the data to correlate haemochromatosis to the data.
Haemochromatosis is a genetic disorder where individuals are unable to get rid of excess iron. If properly diagnosed, this is easily treatable with bloodletting (blood donation in this century :-)
Haemochromatosis is the sort of genetic disease that could really screw with data like this. Undiagnosed haemocromatosis leads to early death....so the correlation of excess iron to longevity should be treated as a weak piece of evidence until we can exclude people with the genetic disorder.
Great comment! For the iron - ageing link, we used two pieces of data: the effect of DNA on blood iron levels, and the effect of DNA on how long and healthy you live. We found the mutation that causes hereditary haemochromatosis had a large effect on early death (as expected). However, other genetic variations that increased blood iron levels also had a proportional effect on early death. So it looks like it is higher iron levels, not just haemochromatosis, that may accelerate the rate of ageing.
"Multivariate MR of iron-related traits on healthspan, parental lifespan, and longevity shows a protective effect for transferrin and a deleterious effect for serum iron."
not in life sciences too, and my interpretation of the above phrase is "if your body iron processing machine works well - good, if not - then bad" which i'd say would be pretty expected, and in general article sounds like there is positive correlation between well working iron related metabolism and longevity which seems not that surprising. They don't seem to address the machinery and nature of that correlation.
Hi there, I'm one of the study authors. Iron plays many roles and is needed for your body to function correctly, but we found people who had higher blood iron levels because of their genes were getting diseases earlier in life and had a shorter lifespan, on average.
Generally better, but the answer is a little more nuanced.
We had access to data describing how DNA influences iron levels, and how DNA influences years of healthy life. This showed that the same DNA variations that change your levels of iron in the blood also change how many years you spend in good health.
Now, these genetic effects are relatively small, so we can confidently say that people who carry genes which cause them to have slightly lower iron levels (i.e. one standard deviation) than the population average will be healthier, on average.
However, extrapolating that further can be dangerous - for example, anyone with iron levels three standard deviations lower will not be three times healthier, they'd have no iron in their blood and be dead.
They say they do, but I'd say they didn't do that properly.
Quoting the relevant part for you that lead me to this conclusion:
> "For one, there are currently no widely accepted standards for measuring healthspan. Zenin et al. define healthspan based on the incidence of the eight most common diseases increasing exponentially in incidence with age in their sample. "
> "Recent estimates suggest the genetic components of both human lifespan (i.e. the number of years lived) and healthspan (the number of years lived in good health free of morbidities) are only around 10%"
> "Parental lifespan correlates strongly with both healthspan (rg = 0.70; SE = 0.04) and longevity (rg = 0.81; SE = 0.08), while healthspan and longevity show a weaker correlation with each other (rg = 0.51; SE = 0.09)"
The authors may pretend they studied healthspan, but I'd rather say they studied biased data (as death is well reported, while autonomy is harder to asses according to their own words), on something poorly correlated with healthspan, which even if we assume perfect correlation can only explain about 10% of the difference.
> healthspan correlated more strongly with metabolic traits (such as type 2 diabetes) than the other studies, and showed negative genetic correlations with depression and cancers, especially melanoma
So I would not call the article rubbish, but I would be careful to extrapolate that iron matters so much:
> These correlation estimates ranged from 0.013 between healthspan and longevity to 0.094 between healthspan and parental lifespan, reflecting a small degree of sample overlap and/or phenotypic correlation
Indeed, parental lifespan may explain more! Even better, look at the coefficient in table 3:
βlongevity is 5x more than βhealthspan , and one coefficient even changes sign (ferritin) possibly due to haemochromatosis (I'm too lazy to check, I've already spent too much time on this article) so I wouldn't jump to conclusions
Rubbish comment. The paper studies many things and found iron metabolism is a common factor for aging.
"Haem synthesis declines with age and its deficiency leads to iron accumulation, oxidative stress, and mitochondrial dysfunction. In turn, iron accumulation helps pathogens to sustain an infection, which is in line with the known increase in infection susceptibility with age. In the brain, abnormal iron homeostasis is commonly seen in neurodegenerative diseases such as Alzheimer’s and Parkinson’s disease and multiple sclerosis"
Also FYI in the title "Multivariate genomic scan implicates novel loci and haem metabolism in human ageing", haem implies iron.
I submitted it yesterday with the original title and it didn't get any attention, possibly because it's a little dense. I'm happy to see the article on the front page now, from a different submitter, as the topic is generally of interest here.
Mendelian randomisation is somewhat of a gold standard for analysing this kind of data. It is a technique to try and establish causal relationships between gene variations and phenotype. Definitely not the kind of crap pushed by a "supplement pusher".
The tl;dr is they found some genes relating to iron metabolism and other key cellular processes that are correlated with longevity and healthy ageing. It doesn't seem particularly actionable unless you're a DNA researcher.
People say this so much that is now used as a way to dismiss any study. It’s almost becoming a meme.
I think is clear that the baseline of any study is that correlation does not imply causation. You start there. That’s a basic intelligent start. You never take for granted that one thing causes the other. The ideas is to try to identify if there are any potential paths for causation.
Otherwise there’s not even a point in sharing or discussing these studies.
I’m not saying that bringing this up is wrong, but does it need to be said so much? I think it’s more interesting to analyze beyond that and try to see if there’s causation. We already know there’s an abysmal amount of correlated things but there must be a percentage of those things which are in fact caused by another at a certain level. I would love these discussions to be more focused on this and not just doing the plain classic dismissal of “correlation doesn’t mean causation”.
> I think is clear that the baseline of any study is that correlation does not imply causation. You start there. That’s a basic intelligent start. You never take for granted that one thing causes the other. The ideas is to try to identify if there are any potential paths for causation.
You might start there, but an incredible number of people read the headline and think "oh man; I guess I gotta increase my iron intake".
It's said a lot because it's so easy to draw a false conclusion from correlations.
It really is almost a meme. You could almost perfectly plug it into the "Learned Sage Points Out That Powerball Not As Much After Taxes" Onion article:
It needs to be said so much because people weaponize scientific studies in the pursuit of their political agenda, media twist headlines to get clicks, etc. I've lost track of the number of articles that bury the "the authors of the study caution against any definitive conclusion about causation/correlation"
If the goal is rhetoric and not truth, no amount of repeating droll truisms is going to prevent that.
Additionally this comment thread has been fairly tame.
It doesn’t need to be said so much, obviously. I’ve counted what, 3 times so far in just this thread?
If this were an observational study, I'd completely agree. In that case, it could be that people who eat better diets and do exercise happen to have lower iron levels, but it is the diet and exercise that makes them live longer, not the iron.
Our study actually used a different method! We found that the genes that raise iron levels in the blood also have an effect on how long and healthy we live. Since you don't get to choose the genes you are born with, this method is more robust to confounding. It's not perfect by any means, but it does provide reasonably robust evidence for a causal role of iron rather than just a correlation. Still, we would like to verify this in follow-up experiments, possibly using animal models.
from the article: "Interestingly, protective variants near FOXO3 are associated with a reduction in metabolic syndrome, but also with a reduction in cognitive ability."
> Twenty one of the 24 multivariate GWAS loci reaching genome-wide significance had directionally consistent effects in the three studied datasets and 18 were nominally significant (P < 0.05) in two or more datasets (Supplementary Fig. 1).
Unrelated to the content of article but why are “twenty one” and “three” and “two” spelled out but 24 and 18 are written as numerals? Is there any rhyme or reason to this? I find this kind of inconsistency so distracting when reading technical papers. It makes me think some of the numbers were edited/changed last-minute in subsequent drafts.
Style rules often look like: spell out any number under twelve (sometimes twenty), and spell out any number that starts a sentence. If you're unlucky, that can get you sentences like the above.
What does your style guide say about periods? I notice none of your bullets ended with a period. As a programmer I often struggle with whether I should end my comment with a period.
First, I dislike the AP style guide. So my own writing will sometimes violate it just to please me.
As for periods, any phrase that is a complete sentence must have a final punctuation mark. So my bullet points would've required periods in a report. They also could've been connected by semicolons with a conjuction at the end of the second-to-last point:
- A;
- B; or
- C.
The idea is to stick to proper English sentences. The bullet points are just incidental formatting.
I'm not sure about other fields, but in Neuroscience we were taught to follow APA guidelines in our undergrad research fundamentals class. The APA style guide is very specific about things like when to spell out numbers, etc.
Seems to be a style that doesn't add anything useful.
In related news, a formal convention that really grates is the american way of spelling out a number then putting it in brackets after: "You are permitted fourteen (14) days holiday a year" kind of thing. If anyone can shed any light on that I'd be interested.
Oh I actually might know the answer to that one! According to my Commercial Paper professor it was an early way to prevent fraud and mutual mistake. It’s fairly easy to sneak an extra zero on the end of a written number or to transpose two numbers without anyone noticing. Checks still do this: “Pay twenty and xx/100 dollars” followed by “$20.00”
In theory if the written number doesn’t match the numeral the teller shouldn’t accept the check.
If only there were a section of scientific papers devoted to explaining the Methods used to obtain the Results. It’d be especially great if it came after an Introduction explaining the context and was followed by a Discussion of the merits, weaknesses, and impact of the study.
Hey friend. You're 100% right for pointing out that this stuff is addressed in the paper. I think the sarcasm is a disservice to everyone, both you and the parent comment included. Further, he may not have had (or may not have) the time to fully analyze the paper, and may be asking if others are aware if that specific concern is addressed, which seems like a reasonable comment.
Sorry, I don't really agree. HN is full of people who react to headlines without reading the article and the underlying materials, and in my opinion it greatly decreases the quality of discussion.
I was reading ..somewhere.. that a person having a negative interaction on an online forum increased the likelihood of that person then going on to make a negative comment themselves 300%. When I read that HN's zealous downvotes of slightly impolite comments started to make a bit more sense to me. It's better to avoid snark and sarcasm, as fun as they are.
If your life revolves around acceptance by HN mods, then you need to wakeup from The Matrix!
Nobody wants another cesspool like /. was in its hay day, but ultimately only HackerNews/Ycombinator are harmed by this phenom of over-moderation.
For example, whose brilliant idea was it to create a user community that downvotes on comments hostile to Chinese Communist Party (that aims nuclear weapons at us while supporting North Korea in every sense and planning to kill our trading partners in Hong Kong and Taiwan) or on comments brutally critical of Apple for abusive trade practices? A lot of the downvotes also come from users that are obviously very inexperienced and unable to understand subtle contexts leading one to believe that a great way to solve this “problem” would be to only allow moderation by senior members of the community, as determined by peer votes, not by account age.
Isn't 500 karma needed to downvote? Karma might be an indicator of seniority indicated by peer vote. As for the other issues, there are ways of being highly critical without needing to be brutal.
You are also 100% right that I could have been more positive and constructive. I did intend to be lighthearted rather than mean, but tone is difficult in plaintext.
I see a lot of these types of comments on hn/reddit/twitter/etc, and (painting with a broad brush) the authors are typically pointing out trivial or spurious concerns, and phrasing it to cast doubt on the entire analysis. Often, the authors are asking about something that's easily searchable (here, cmd-F for "sex" will show a subsection of the methods dedicated to sex-specific effects), or naively ignoring the existing practices of the field.
There are no bad questions, but there are certainly bad attitudes. I'll be the first to admit the flaws in academia, including that many papers are imperfect and many are bad, but I think the dismissive comments from authors who haven't engaged with the material are just as much a disserve to the conversation around any study.
Of course, that may be another issue with tone in plaintext :)
Great and important question! All data we used was corrected for sex. In other words, we knew the effect of DNA on iron levels, and the effect of DNA on lifespan/healthspan, regardless of the person's sex.
This is a good question to ask and I am not sure why you are being downvoted. The answer is that p-hacking was a real problem back when these kind of large multi-gene frequency analysis studies were just starting out. Not intentionally, mind you, just that the people running the studies were biolgists first, and statisticians only on an ad hoc basis. In the late 90s and early 2000s there was a lot of debate about the poor quality of statistical analysis in the field and things started to improve. Today these studies have a bunch of descriptivist and bayesian tools they can use to sort out real positives and negatives from false ones. The result is that these studies now discover real and important genes that impact complex diseases. We are finally getting close to the dream of personalised medicine.
> The effects of loci of interest on male and female lifespan are largely the same, although their effect on survival may be slightly stronger in middle age (40–60 years) compared to old age (>80 years).
This is interesting. I have a feeling that because there are many congenital cardiovascular and other congenital diseases that one usually succumbs to in the 40-60 range that is what we are seeing a stronger link to genes here. Unless a parents passes away before the age of 55 from a heart attack it's not considered hereditary.
I find it interesting when studied like this are performed. One of the best predictors of life expectancy in elderly individuals is grip strength; not because of the fact that one has a strong grip, but because a stronger grip means more muscle mass leading to longer life. When I see that iron levels are correlated to longer lifespans I wonder if there's something else going on. Many women, in particular, walk around with iron deficient anemia and don't realize their fatigue is because of their lack of iron. Anyway, I'm excited for this line of research to be further delineated.
75 comments
[ 3.6 ms ] story [ 142 ms ] threadEdit: specified as negatively correlated
Haemochromatosis is a genetic disorder where individuals are unable to get rid of excess iron. If properly diagnosed, this is easily treatable with bloodletting (blood donation in this century :-)
Haemochromatosis is the sort of genetic disease that could really screw with data like this. Undiagnosed haemocromatosis leads to early death....so the correlation of excess iron to longevity should be treated as a weak piece of evidence until we can exclude people with the genetic disorder.
No, it's saying neither. The submission title could be seen as misleading.
not in life sciences too, and my interpretation of the above phrase is "if your body iron processing machine works well - good, if not - then bad" which i'd say would be pretty expected, and in general article sounds like there is positive correlation between well working iron related metabolism and longevity which seems not that surprising. They don't seem to address the machinery and nature of that correlation.
We had access to data describing how DNA influences iron levels, and how DNA influences years of healthy life. This showed that the same DNA variations that change your levels of iron in the blood also change how many years you spend in good health.
Now, these genetic effects are relatively small, so we can confidently say that people who carry genes which cause them to have slightly lower iron levels (i.e. one standard deviation) than the population average will be healthier, on average.
However, extrapolating that further can be dangerous - for example, anyone with iron levels three standard deviations lower will not be three times healthier, they'd have no iron in their blood and be dead.
Frailty is another metric, probably more important as you want to live a long live while being able to use your days.
Quoting the relevant part for you that lead me to this conclusion:
> "For one, there are currently no widely accepted standards for measuring healthspan. Zenin et al. define healthspan based on the incidence of the eight most common diseases increasing exponentially in incidence with age in their sample. "
> "Recent estimates suggest the genetic components of both human lifespan (i.e. the number of years lived) and healthspan (the number of years lived in good health free of morbidities) are only around 10%"
> "Parental lifespan correlates strongly with both healthspan (rg = 0.70; SE = 0.04) and longevity (rg = 0.81; SE = 0.08), while healthspan and longevity show a weaker correlation with each other (rg = 0.51; SE = 0.09)"
The authors may pretend they studied healthspan, but I'd rather say they studied biased data (as death is well reported, while autonomy is harder to asses according to their own words), on something poorly correlated with healthspan, which even if we assume perfect correlation can only explain about 10% of the difference.
> healthspan correlated more strongly with metabolic traits (such as type 2 diabetes) than the other studies, and showed negative genetic correlations with depression and cancers, especially melanoma
So I would not call the article rubbish, but I would be careful to extrapolate that iron matters so much:
> These correlation estimates ranged from 0.013 between healthspan and longevity to 0.094 between healthspan and parental lifespan, reflecting a small degree of sample overlap and/or phenotypic correlation
Indeed, parental lifespan may explain more! Even better, look at the coefficient in table 3:
> https://www.nature.com/articles/s41467-020-17312-3/tables/3
βlongevity is 5x more than βhealthspan , and one coefficient even changes sign (ferritin) possibly due to haemochromatosis (I'm too lazy to check, I've already spent too much time on this article) so I wouldn't jump to conclusions
Later on we get: "Mendelian randomisation of iron traits" and the HN title looks very suspect. The sort of cobblers a supplement pusher might write.
"Haem synthesis declines with age and its deficiency leads to iron accumulation, oxidative stress, and mitochondrial dysfunction. In turn, iron accumulation helps pathogens to sustain an infection, which is in line with the known increase in infection susceptibility with age. In the brain, abnormal iron homeostasis is commonly seen in neurodegenerative diseases such as Alzheimer’s and Parkinson’s disease and multiple sclerosis"
Also FYI in the title "Multivariate genomic scan implicates novel loci and haem metabolism in human ageing", haem implies iron.
>What does haem mean in biology?
haem. 1. (Science: biochemistry) compounds of iron complexed in a porphyrin (tetrapyrrole) ring that differ in side chain composition.
I read it, but didn’t understand what the conclusion is, or if it’s actionable or not. Should I better monitor my iron intake or something?
I think is clear that the baseline of any study is that correlation does not imply causation. You start there. That’s a basic intelligent start. You never take for granted that one thing causes the other. The ideas is to try to identify if there are any potential paths for causation.
Otherwise there’s not even a point in sharing or discussing these studies.
I’m not saying that bringing this up is wrong, but does it need to be said so much? I think it’s more interesting to analyze beyond that and try to see if there’s causation. We already know there’s an abysmal amount of correlated things but there must be a percentage of those things which are in fact caused by another at a certain level. I would love these discussions to be more focused on this and not just doing the plain classic dismissal of “correlation doesn’t mean causation”.
You might start there, but an incredible number of people read the headline and think "oh man; I guess I gotta increase my iron intake".
It's said a lot because it's so easy to draw a false conclusion from correlations.
https://local.theonion.com/learned-sage-points-out-that-powe...
It doesn’t need to be said so much, obviously. I’ve counted what, 3 times so far in just this thread?
It doesn’t make you sound smarter.
Our study actually used a different method! We found that the genes that raise iron levels in the blood also have an effect on how long and healthy we live. Since you don't get to choose the genes you are born with, this method is more robust to confounding. It's not perfect by any means, but it does provide reasonably robust evidence for a causal role of iron rather than just a correlation. Still, we would like to verify this in follow-up experiments, possibly using animal models.
Unrelated to the content of article but why are “twenty one” and “three” and “two” spelled out but 24 and 18 are written as numerals? Is there any rhyme or reason to this? I find this kind of inconsistency so distracting when reading technical papers. It makes me think some of the numbers were edited/changed last-minute in subsequent drafts.
Edit: reading the other comments, beginning vs non-beginning of a sentence also make sense.
- Numbers up to ten are written as words
- Numbers 11 and higher are numerals
- Any number that starts a sentence is written as a word
As for periods, any phrase that is a complete sentence must have a final punctuation mark. So my bullet points would've required periods in a report. They also could've been connected by semicolons with a conjuction at the end of the second-to-last point:
- A;
- B; or
- C.
The idea is to stick to proper English sentences. The bullet points are just incidental formatting.
In related news, a formal convention that really grates is the american way of spelling out a number then putting it in brackets after: "You are permitted fourteen (14) days holiday a year" kind of thing. If anyone can shed any light on that I'd be interested.
In theory if the written number doesn’t match the numeral the teller shouldn’t accept the check.
Nobody wants another cesspool like /. was in its hay day, but ultimately only HackerNews/Ycombinator are harmed by this phenom of over-moderation.
For example, whose brilliant idea was it to create a user community that downvotes on comments hostile to Chinese Communist Party (that aims nuclear weapons at us while supporting North Korea in every sense and planning to kill our trading partners in Hong Kong and Taiwan) or on comments brutally critical of Apple for abusive trade practices? A lot of the downvotes also come from users that are obviously very inexperienced and unable to understand subtle contexts leading one to believe that a great way to solve this “problem” would be to only allow moderation by senior members of the community, as determined by peer votes, not by account age.
You are also 100% right that I could have been more positive and constructive. I did intend to be lighthearted rather than mean, but tone is difficult in plaintext.
I see a lot of these types of comments on hn/reddit/twitter/etc, and (painting with a broad brush) the authors are typically pointing out trivial or spurious concerns, and phrasing it to cast doubt on the entire analysis. Often, the authors are asking about something that's easily searchable (here, cmd-F for "sex" will show a subsection of the methods dedicated to sex-specific effects), or naively ignoring the existing practices of the field.
There are no bad questions, but there are certainly bad attitudes. I'll be the first to admit the flaws in academia, including that many papers are imperfect and many are bad, but I think the dismissive comments from authors who haven't engaged with the material are just as much a disserve to the conversation around any study.
Of course, that may be another issue with tone in plaintext :)
A good further read is the 1998 paper that describes the problem in detail: http://ib.berkeley.edu/labs/slatkin/popgenjclub/pdf/terwilli...
Title is: "Multivariate genomic scan implicates novel loci and haem metabolism in human ageing".
This is interesting. I have a feeling that because there are many congenital cardiovascular and other congenital diseases that one usually succumbs to in the 40-60 range that is what we are seeing a stronger link to genes here. Unless a parents passes away before the age of 55 from a heart attack it's not considered hereditary.
I find it interesting when studied like this are performed. One of the best predictors of life expectancy in elderly individuals is grip strength; not because of the fact that one has a strong grip, but because a stronger grip means more muscle mass leading to longer life. When I see that iron levels are correlated to longer lifespans I wonder if there's something else going on. Many women, in particular, walk around with iron deficient anemia and don't realize their fatigue is because of their lack of iron. Anyway, I'm excited for this line of research to be further delineated.