Read on Twitter
Tons of chatter about the new prescription omega 3 data coming forth. Lots of takes about the study designs, n3 formulations, differences across populations, variation in comparator groups. All important ?- but I think there's a bigger 
in the room. #cardiotwitter #lipids #AHA20
in the room. #cardiotwitter #lipids #AHA20
Intervening with n3 fatty acids is unlike other drugs bc they are a nutrient. There is no placebo for n3s in the sense that:
1) you have questions about what is the ideal comparator; typically isocaloric oil.
2) you don't have an unexposed group - there are endogenous levels.
1) you have questions about what is the ideal comparator; typically isocaloric oil.
2) you don't have an unexposed group - there are endogenous levels.
There are numerous reasons for n3 enthusiasm, but playing by evidence hierarchy, the top reason is that meta-analyses of cohorts demonstrate that higher blood levels of n3s result in reduced risk of fatal CHD https://jamanetwork.com/journals/jamainternalmedicine/article-abstract/2530286
This literature is messy, in large part because you can measure omega 3s in different blood lipid fractions. The best one to use is erythrocyte Omega 3 Index (n3s in red blood cell membranes as % total fatty acids). The cohorts have been harmonized to RBC n3 index:
So the observational epidemiology with biomarker data broadly points towards a goal of getting 8% RBC N3Index. Where are the US and Canada at? Not so hot. https://www.sciencedirect.com/science/article/pii/S0163782715300333#f0010
In NHANES 2003-2004 (admittedly a bit dated now), I used some of the harmonization equations to estimate RBC n3 index from plasma and it puts us at about 2-3% RBC n3 Index: https://www.mdpi.com/2072-6643/7/12/5534/htm
So summary: status markers are associated with CVD risk. Population has low status.
Not wild to intervene right? But how do you intervene?
Not wild to intervene right? But how do you intervene?
The ideal way to do such trials would be to recruit individuals in the risk range (<4% RBC n3 Index) and randomize to achieving a >8% RBC n3 Index. That takes more time and money than a typical trial, so trialists tend to recruit a population & randomized to single dose vs compar
a key thing to think abt w this design is that the hypothesis tested is not necessarily the one we care about (does going from 4-8% RBC n3Index reduce risk) but ends up testing does shifting the baseline status distribution reduce risk? Depends on baseline status & other factors.
So what do the trials that we're all debating about actually look like? I summarized recent trials & some historical ones below. specifics in the table but reflections below:
1) Few trials are measuring RBC N3 Index at baseline and at follow-up time points. This leaves us having to try to impute the RBC n3 index from plasma measures. Some studies report plasma absolute n3 levels but not total fatty acid levels, so you can't assess % total fatty acids
2) few trials seem to capture folks in the <4% n3 index range which appears consistently in the literature. Is the literature wrong? or are the trials recruiting participants that majority don't look like the rest of the population (e.g. consuming 
fish, using supplements)
fish, using supplements)
3) None of these trials are reporting formal interactions between supplementation and baseline status. Even then this is a bit challenging to handle statistically because considering baseline status doesn't mean you've achieved cardioprotective status indicators at follow up.
4) The trials are so wildly heterogeneous, that folks thinking they can derive whether its the formulation (@.EPA lovers) or Met risk factor differences driving the effects aren't really able to discern that from the way the trials are reported & analyzed. Its fun to guess tho
5) This literature probably won't see clarity until:
a) we get a group together with money to measure baseline & achieved RBC status in these trials and do an individual participant meta analysis to try to make some sense of whether these trials tested what we care about
a) we get a group together with money to measure baseline & achieved RBC status in these trials and do an individual participant meta analysis to try to make some sense of whether these trials tested what we care about
b) rather than continuing to shoot into the dark with single dose vs comparator large endpoint trials, more nuanced, adequate trials that recruit based on RBC n3 index cut points and randomize to achieving and stabilizing at a specific RBC n3 index dose are needed.
That last point applies to most things in nutrition. Nutrients aren't drugs. We have homeostatic systems that aim to retain nutrients in our bodies. Baseline status in almost every physiology trial predicts response 2 interventions. We can't just ignore that (hi Vitamin D trials)
It's quite disconcerting to watch the tackling of nutrient- chronic dz endpoint rlx in serious hard endpoint trials & seeing them flop on addressing the basics of nutrition science. N3s and VitD are easily supplementable nutrients w decent status markers &we're flubbing these.
