Cholesterol Manipulation in 10 Keto Women

Good morning, everyone. I'm very excited for this paper review video, which is going to be going over data that colleagues and I just published in Frontiers in Endocrinology. The title of the paper is Thyroid Markers and Body Composition Predict LDL Cholesterol Change in Lean Healthy Women on a Ketogenic Diet: Experimental Support for the Lipid Energy Model by co-first authors Isabella Cooper and Claudio Sanchez Pizarro, followed by myself, Nick Norwitz, Dave Feldman, and our senior and anchoring author is our colleague Adrienne Soto-Motta. So, to dig right into it, this study was an interventional trial where Dr. Cooper, as part of her PhD project, took lean, healthy women, 10 lean, healthy women, all BMI below 25, mean BMI was actually 20.5, so quite lean, who interestingly had been keto-adapted for quite a long time. They had just chosen to adopt a ketogenic diet. These were the participants selected for. Recruitment criteria required they had been keto for at least 6 months confirmed, although mean time in ketosis was actually 3.9 years. And for this trial, what they did interestingly was the first phase was continued nutritional ketosis. So they measured variables for 3 weeks while the women maintained basically their habitual diet with nutritional ketosis at least 0.5 millimoles or more. So they maintained their ketogenic diet. Then in the middle phase, phase 2, is when they suppress ketosis— suppression of ketosis, S-U-K. And to do this, they adopted the UK, like, healthy eating guidelines. Healthy eating, the quotes mean carbs. Basically, the women started eating more carbs. About 55% of kilocalories from carbs were the instructions to suppress ketosis, which was confirmed as ketone levels below 0.3 millivolts. Now, I will say that throughout the trial, the women were incredibly rigorous in their adherence to the protocol. And all women maintained nutritional ketosis during the nutritional ketosis phase, suppressed it during the suppression of ketosis phase. And then phase 3 was to go back to their habitual ketogenic diet. So then they came back into nutritional ketosis. So phase 1, keto diet. Phase 2, you suppress ketosis with carbs. Phase 3, you go back into ketosis. And throughout these phases, you can measure changes in hormones and other variables to see, you know, what is changing in these keto-adapted people as you add back carbs and then remove the carbs again. Again. And the focus of this particular paper, while there will be many coming out of this trial, was changes in cholesterol, specifically LDL cholesterol, and what predicts those changes, the emphasis being body mass index and thyroid hormone. So to focus on the former, body mass index, we've shown in prior trials, particularly a cohort study, that there is an inverse association between BMI, body, body mass index, and LDL cholesterol change, such that leaner people tend to have higher LDL while on keto, um, and larger increases in LDL when going keto. Now, a converse of this would be that when you add back carbs to people who are lean and on keto with elevated LDL, those people should also see the biggest drops because they're presumably the ones who had the biggest increases to begin with. So basically, their flux in LDL is the largest, which can be tested in this interventional trial because you had these women who, although all lean, did have a spectrum of leanness, and you added back the carbs. So this is an experimental test for the lipid energy model. And you can see that test bear out here in Figure 2, which I think is really beautiful. So to unpack it for a minute, what you're looking at here on the y-axis on the left is LDL levels going from, I mean, just under 100 to over 250. And then there's a key where there's a color code for BMI for the different 10 participants, whereby red is higher BMI closer to 25, although all these women again were lean. And then the blue-purple is, um, lower BMI. And so what you'd expect according to the lipid energy model and prior predictions, prior data saying leaner people will have higher LDL while keto, is that in phase 1 nutritional ketosis, the people with lower BMI, the blue-purple, should be higher in terms of LDL. And indeed they are. Just looking at, um, part 1, phase 1, for a second, you see that there's more red, higher BMI, lower down, and more blue-purple higher up. Consistent with lipid energy model. Then what do you see in phase 2 when they add back carbs? You see the lean people, the people that started with a higher LDL, the blue and purple, tend to see the largest drops and then going, you know, trending back towards what they were in ketosis when they go to phase 3. So this is just experimental confirmation of prior predictions and prior data with respect to the lipid energy model. The leaner people have higher LDL when keto and larger drops when you add back carbs. So that's pretty cool. Now, it was further tested experimentally in multivariate models to see, you know, is BMI a robust predictor of LDL change along with other markers of body composition, which we actually did have in this study. And indeed, all like body anthropomorphic metrics were predictors of LDL change. So again, consistent with the lipid-energy model. Another thing we did in the study was compare, you know, different covariates that may affect LDL to BMI. And to explain that and unpack that a little bit, I, I'll just read a section from the paper that I think is very important, which was, as shown in the sensitivity analysis in Table 3, BMI remained a robust and significant predictor of LDL change despite the addition of any covariate, and no covariate remains significant after accounting for the effect of BMI. Basically what this is saying is BMI dominates. When you add other things into the model, BMI remains significant. But if you add BMI to other things, those things become insignificant typically. So BMI is the dominant variable in determining LDL change. Then Dr. Sotomayor did something really brilliant, which is a curious analysis where he modeled— well, what if these women, say, were consuming basically most of their calories, almost all their calories from saturated fat, which would be really hard to do in the real world. But he modeled it such that if saturated fat intake was as high as 90% of total energy So that would be eating a 4:1 therapeutic ketogenic diet where every single calorie from saturated fat, or from fat, was saturated fat. So you would have to be like guzzling coconut oil and then eating a little bit of pure protein basically to achieve this. This would be basically physiologically impossible. But even if that were the case, saturated fat intake as high as 90% of total energy intake, if that were the case, BMI would still remain a significant and virtually unchanged predictor of LDL. So bottom line takeaway point Here is that BMI dominates over saturated fat. Now, um, I'm just going to put a pin in this one because I will tease— we have even more robust data really proving this point and probably putting the nail on the coffin of the idea that saturated fat drives LDL in lean people on ketogenic diets. But let's put that aside for now. Bottom line, BMI is a dominant factor in determining LDL rises when you go keto and LDL drops when you add back carbs. Okay, BMI is dominant. But what this study also adds to the literature is another piece of the puzzle, which is thyroid hormone. So the lipid energy model has to do with how the body responds to carbohydrate restriction in lean people, typically, um, to, you know, meet energy demands. All about energy. And thyroid hormone is a major determinant of energy flux throughout the body. So we would actually think that thyroid hormone would have an important predictive effect on LDL in addition to BMI, that they might be independent and interacting predictors, which we had sufficient data to test in this study. And what did we find? We indeed found that free active thyroid hormone, free T3, was, um, an independent predictor in addition to BMI for LDL changes. And you can see that here really beautifully in the multi-dimensional predictive performance model. Um, what you want— I want you to look at first is the green polygon, the free T3. So the way this diagram works is better predictive models for LDL will have a larger surface area. So you see the green has a pretty big surface area. It's smaller than BMI, the blue. So BMI is still better than thyroid, but they're each independent predictors. But what happens when you combine them? That you see in the brown polygon, the BMI and the free T3. The combination of these two independent predictors of LDL change provides the best prediction, which is really cool. And one thing this, you know, study adds to the literature. Now, interpretation of the thyroid change is actually a little bit complicated because what you see, consistent with other data, is that when you add back carbs in phase 2, the free T3 does go up. Now, higher thyroid hormone tends to associate with higher energy flux, um, and lower thyroid hormone tends to associate with lower energy flux. But with macronutrient shifts, things get a little bit more complicated. So, you know, one might predict, oh, on keto the free T3 was lower, so the metabolic rate and energy flux will be lower. Well, no, that's not necessarily the case. You can actually see that in this study and in other studies, even though on keto free T3 is lower, the basal metabolic rate isn't changed. And while there are some conditions, it's like called sick euthyroid syndrome, associated with, say, anorexia or critical illness, where the TSH can remain normal but the free T3 drops, and that's kind of an indication of, um, critical illness, that's very unlikely in this case, again, because these were healthy women with normal basal metabolic rates. So how do you interpret this flux in T3? I don't think it's really clear. My personal opinion is it probably has to do something with thyroid thyroid hormone sensitivity, analogous to things like insulin sensitivity and leptin sensitivity, where the level of the hormone itself doesn't necessarily reflect, um, activity of the hormone's axis. So for example, in hyperinsulinemia, in like prediabetes, the insulin level is extremely high, but insulin signaling isn't necessarily high in correspondence with the high insulin because there's resistance. So, you know, there's some data to suggest that when you lower carbs, your free T3 might go down, but it doesn't mean you have less thyroid axis activity. It's just that you're more sensitive to the thyroid hormone, so you don't need as much. Nevertheless, what we're showing in this study is that measuring free T3 is an independent predictor of LDL change. Exactly the pathway of causality isn't clear, and that will be a topic for future research. But anyway, I'm really proud of this team, especially Dr. Cooper, and these data. So just So to just summarize again, what this is showing in an interventional trial is that yes, consistent with prior data, BMI is a major determinant, a major independent determinant of LDL change. It dominates over other determinants. It dominates over saturated fat for sure. Again, teaser for future data to come. But these data also show that thyroid hormone, free T3, is an independent predictor, and that combining the two provides the best prediction for LDL change, advancing the lipid energy model and evolving it. So, um, congratulations to my team. I look forward to the discussions about this paper and those to come. Have a lovely day.