What to Make of Vitamin D, MS, & Mendelian Randomization – MedPage Today

A Mendelian randomization study suggested that low vitamin D levels may play a causal role in the development of multiple sclerosis.

In an analysis of data from large cohort studies, each genetically determined standard deviation drop in log-transformed 25-hydroxyvitamin D (25OHD) levels conferred a two-fold increase in the odds of having MS, Brent Richards, MD, of McGill University, and colleagues reported online in PLOS Medicine.

The point of Mendelian randomization is to avoid the confounding that typically occurs with observational data. It is supposed to be a way to help discern whether some environmental factor — like vitamin D levels — is causally linked to an outcome.

So instead of looking at the relationship between MS and measured vitamin D levels — an environmental factor that can be influenced by confounding — this strategy looks at gene variants that aren’t impacted by those limitations to get a sense of the direct relationship between the risk factor and outcome.

“It’s called ‘nature’s randomized controlled trial’ because whether someone gets a certain variant is a random event,” Richards said, adding that his group is “not claiming causality here. Rather, we’re providing evidence that supports causality.”

But how accurate is Mendelian randomization for discerning causality?

“It depends on who you ask,” Richards told MedPage Today. “I believe it can produce evidence supporting or contradicting causality.”

He thinks it’s especially useful in cases where it’s unlikely that a randomized controlled trial of necessary length and scope would ever be accomplished — as would be the case with vitamin D, he said.

For their study, Richards and colleagues looked at data from the largest genome-wide association study (GWAS) to date on vitamin D — the SUNLIGHT study of nearly 34,000 people.

That study revealed four major single nucleotide polymorphisms (SNPs) that were linked with decreased 25OHD levels.

They then validated those four SNPs using data on 2,347 participants in population-based cohort study, the Canadian Multicenter Osteoporosis Study — and confirmed that the count of low vitamin D alleles was strongly associated with lower levels of 25OHD.

Finally they conducted the Mendelian randomization analysis using data from the International Multiple Sclerosis Genetics Consortium study, the largest GWAS to date for MS — totaling nearly 14,500 cases and 24,000 healthy controls.

They found that each genetically determined standard deviation drop in log-transformed 25OHD levels conferred a two-fold increase in the odds of MS (95% CI 1.7 to 2.5, P=7.72 x 10-12).

That relationship persisted in sensitivity analyses that excluded SNPs possibly influenced by population stratification or pleiotropy and included only SNPs involved in 25OHD synthesis or metabolism.

They said the findings are consistent with observational evidence that low vitamin D levels influence the risk of MS — including epidemiological studies that show a higher prevalence of MS in higher-latitude regions with diminished exposure to sunlight.

The next step, Richards said, would be to do a clinical trial to investigate whether vitamin D supplementation might prevent or delay the onset of MS in patients who are already at elevated risk, such as those having family members with the condition.

“We think these findings strongly suggest people at risk for MS should ensure they have normal vitamin D levels,” Richards said.

When you ask experts in epidemiology and biostatistics who have worked extensively with Mendelian randomization, they say the study was well executed.

Nicholas Timpson, PhD, an epidemiologist at the University of Bristol, said that Mendelian randomization studies “should not be taken as gold standard as there are a series of caveats which, as for most studies, may influence interpretation” — but he called the findings “compelling evidence.”

“It is well undertaken and the use of Mendelian randomization in this case is a genuine contribution to the potential for a causal relationship between exposure and disease,” Timpson told MedPage Today.

Stephen Burgess, PhD, a biostatistician at the University of Cambridge, noted that one strength of the current study is that it doesn’t depend on a single genetic variant, but looks at four individual variants that contribute to vitamin D status, making for “strong support of the claim of causation.”

But a limitation is that the study is “only able to make the general conclusion that increasing vitamin D levels should lower MS risk,” Burgess told MedPage Today. It gives no indication of what levels of vitamin D are likely to be beneficial or optimal, nor does it suggest which patients vitamin D supplementation would be most beneficial for, he said.

While these questions need to be answered in an RCT, Burgess said there are aspects of Mendelian randomization design that RCTs wouldn’t be able to take into account — and life-long vitamin D exposure is one of them.

“At the end of the day, the proof of the pudding is in the eating,” Burgess said. “What will convince clinicians is this: do results from Mendelian randomization correspond favorably with results from RCTs? At the moment the answer is a resounding yes — statins, PSCK9 inhibitors (positive), and CETP-inhibitors (null) for CHD risk, folic acid for stroke risk (null in Western populations, positive in East Asians) — [are] now supported by trial evidence.”

But what happens when you ask a clinician? Perry Wilson, MD, of Yale, who is a MedPage Today physician reviewer, said the main study outcome — the standard deviation changes in log-transformed 25OHD levels — is “statistically clean, but clinically meaningless.”

“The big finding is that, yes, people with MS were more likely to carry at least one of those low vitamin D genes,” Wilson said. “But beyond saying that, the effect size is very hard to relate, since they were predicated on a standard deviation increase in log-transformed vitamin D.”

Richards disclosed no financial relationships with industry.


Write a Reply or Comment:

Your email address will not be published.*