3 ms·

Thank you for disagreeing. The article was actually so sparse with the information that I misunderstood it for an epidemiological association. I went and found

Thank you for disagreeing. The article was actually so sparse with the information that I misunderstood it for an epidemiological association. I went and found the study in pubmed, and I see that it's actually stronger than that - it's a cohort study with an intervention. It's not gold-standard, randomized-trial-level evidence, but it's still important. I'm still a bit skeptical, because the human genetic data, so far, doesn't back up the association between Vit D deficiency genes and T1D. Nevertheless, I'll revise my statement to agree that both (a) diabetes can cause a decrease in the hormonally active form of vitamin D, and (b) high-quality Grade B evidence suggests that the converse is also true. Nevertheless, I don't see Vitamin D deficiency as the unifying diagnosis here. One could even speculate that the Vit D deficiency was secondary, and the T1D was tertiary, but we really don't have enough information to speculate here and I'll stop now. I probably shouldn't have in the first place.

kragen 16y ago
If 75% of the US population is really vitamin-D-deficient, it's unlikely that most vitamin D deficiency is genetically-caused.
- carbocation 16y ago
  Your point is well-taken, although it's one step removed from the point I'm getting at. My point, it's true, assumes that more than 0% of vitamin D deficiency is genetic. In other words, my point only stands if the heritability of Vit D levels is greater than 0. From that assumption, I discuss the association between Vit D genetics and T1D. (The heritability of Vit D levels has actually been measured, and has been found to be very high - greater than 60% in some populations - so genetic factors are driving a huge amount of the variation that we see.) If there were an association between T1D and Vit D, you would expect an excess of Vit-D-associated variants among those with T1D when compared to matched individuals. It is true that if 75% of the population is Vit D deficient and most of this is due to environmental effects, this would would hurt your power, requiring a larger study before convincing yourself that the null hypothesis really should not be rejected. Hence the couching of my language with "so far," since I know that associations can definitely be uncovered as sample size increases. But given the high heritability, at least in some populations, we shouldn't need a stratospheric sample size to detect association.
  - kragen 16y ago
    How much of the observed vitamin-D deficiency is caused by known Vit-D-associated genetic variants? You can measure heritability without knowing what alleles in particular code a trait, right?
    - carbocation 16y ago
      Correct, you can measure heritability without any genetic information whatsoever. Because of this we can know that, say, 60% of Vit D variability is heritable. I hadn't seen a GWAS on Vitamin D levels before you asked your question, but I pubmedded it and it turns out that a small one came out this month: http://hmg.oxfordjournals.org/cgi/content/full/19/13/2739?view=long&pmid=20418485 http://hmg.oxfordjournals.org/cgi/content/full/19/13/2739?vi... In short, 4 base pairs (out of 3 billion) account for 2.8% of the population variance of circulating Vitamin D. With such a small # of people, it's not surprising that few independent SNPs hit the genome-wide significance level, but it's still pretty impressive that the 4 SNPs that they found can account for so much of the variance. (In other words, it looks like they identified the source of ~5% of the genetic variance, or 2.8% of the total variance.) This looks much like the lipids story: early GWASs give you a handful of loci with influence, and later GWASs with may more participants will give you a much richer set of loci that collectively explain 20-30% of the variance. And keep in mind that these are just the common SNPs.
      - kragen 16y ago
        It would be more than surprising, then, if the human genetic data so far did back up the association between vitamin-D-deficiency genes and T1D.

kragen 16y ago

If 75% of the US population is really vitamin-D-deficient, it's unlikely that most vitamin D deficiency is genetically-caused.

carbocation 16y ago
Your point is well-taken, although it's one step removed from the point I'm getting at. My point, it's true, assumes that more than 0% of vitamin D deficiency is genetic. In other words, my point only stands if the heritability of Vit D levels is greater than 0. From that assumption, I discuss the association between Vit D genetics and T1D. (The heritability of Vit D levels has actually been measured, and has been found to be very high - greater than 60% in some populations - so genetic factors are driving a huge amount of the variation that we see.) If there were an association between T1D and Vit D, you would expect an excess of Vit-D-associated variants among those with T1D when compared to matched individuals. It is true that if 75% of the population is Vit D deficient and most of this is due to environmental effects, this would would hurt your power, requiring a larger study before convincing yourself that the null hypothesis really should not be rejected. Hence the couching of my language with "so far," since I know that associations can definitely be uncovered as sample size increases. But given the high heritability, at least in some populations, we shouldn't need a stratospheric sample size to detect association.
- kragen 16y ago
  How much of the observed vitamin-D deficiency is caused by known Vit-D-associated genetic variants? You can measure heritability without knowing what alleles in particular code a trait, right?
  - carbocation 16y ago
    Correct, you can measure heritability without any genetic information whatsoever. Because of this we can know that, say, 60% of Vit D variability is heritable. I hadn't seen a GWAS on Vitamin D levels before you asked your question, but I pubmedded it and it turns out that a small one came out this month: http://hmg.oxfordjournals.org/cgi/content/full/19/13/2739?view=long&pmid=20418485 http://hmg.oxfordjournals.org/cgi/content/full/19/13/2739?vi... In short, 4 base pairs (out of 3 billion) account for 2.8% of the population variance of circulating Vitamin D. With such a small # of people, it's not surprising that few independent SNPs hit the genome-wide significance level, but it's still pretty impressive that the 4 SNPs that they found can account for so much of the variance. (In other words, it looks like they identified the source of ~5% of the genetic variance, or 2.8% of the total variance.) This looks much like the lipids story: early GWASs give you a handful of loci with influence, and later GWASs with may more participants will give you a much richer set of loci that collectively explain 20-30% of the variance. And keep in mind that these are just the common SNPs.
    - kragen 16y ago
      It would be more than surprising, then, if the human genetic data so far did back up the association between vitamin-D-deficiency genes and T1D.