The Harvard Cookie Study That Broke Cholesterol Science

What if I told you that this cookie lowered my cholesterol more than a multi-billion dollar drug? Oreo! Oreo! Oreo McFlurry! Oreo! Oreo! Oreo! Oreo! Well, that's what I set out to test on an experiment I conducted on myself while I was a student at Harvard Medical School. And spoiler alert, it worked. But not only did it work, the Oreo cookie was more than twice as powerful as high-dose statin medication at lowering my cholesterol. And it worked in 1/3 the time. I put my body on the line in a rigorous 5-month controlled experiment that was eventually published in peer review to figure out the answer to the question: what if everything we've been told about cholesterol is wrong? Stick around and find out, because we're going to talk about just how the cholesterol cookie crumbles. We all know Oreo cookies. Your body flips rapidly. Fat needs a delivery truck, and that truck includes— it worked unbelievably well. This is very different from the general population. What do you think this means? This experiment would provoke a cognitive dissonance. Now to get to the true and honest why behind why I did the Oreo versus statin study in the first place, we first need to go back 6 and a half years to June 1st, 2019. Indulge me for a moment while I share a bit of my personal story. That was the day my life changed. It was the day I went on a ketogenic diet for the first time. Now, at the time, I was spending a decent proportion of my life in the hospital suffering with debilitating ulcerative colitis, a form of inflammatory bowel disease. And while my life held a lot of promise for me on paper, a lot of opportunity, I was a PhD student at Oxford with a position secured at Harvard Medical School, my actual future was very uncertain and far from secure since I just couldn't stay healthy enough to engage with life in any meaningful capacity. I really had no life. I was physically broken and fragile, I was completely socially withdrawn, and I was psychologically just not there. So I did what many patients do in that situation. I got desperate and started trying basically anything, not out of expectation but pure desperation, including a litany of diets. I tried plant-based, vegan, Whole30, specific carbohydrate, low FODMAP, you name it, I tried it. And then on June 1st, 2019, in pure desperation, I tried a ketogenic diet and my life changed. The relentless blood and diarrhea, it relented. My energy and mood improved. My inflammation markers dropped to the lowest they'd been in years. And at my next colonoscopy, I had no signs of ulcerative colitis on biopsy. I was in biopsy-proven remission. But something else happened too. Like so many lean insulin-sensitive people who go low-carb, my LDL "bad cholesterol" grew. It skyrocketed. It did like the Geico gecko transfiguring into Godzilla. It completely transformed. It jumped from the 90s, where it had been my whole life on a standard American diet, to well over 300, and then to over 500, eventually peaking at 574—levels that until recently were thought to only exist in very rare, very terrifying, and very deadly lipid genetic disorders. Then, after several years of study and being terrified, lo and behold, I came to find out that actually I wasn't a unicorn. There are literally thousands of people like me, and this is actually a rather underappreciated, definitely understudied, but becoming more common phenomenon termed a lean mass hyper-responder. Basically, if you're lean with low body fat, generally active and insulin sensitive, and you go low-carb keto, your LDL cholesterol levels and your ApoB have a tendency to explode upward. Now to be clear, and this is important, this is very different from the general population because the general population tends to have overweight, obesity, and be metabolically unhealthy. And in these people, the majority, LDL cholesterol and ApoB tend to not change or even drop on low-carb diets, as colleagues and I, have since shown in the literature, including a meta-analysis of 41 human randomized control trials. So that brings up two massive questions: Why is this occurring? What's the mechanism? And then also, what is the risk? Let's table the risk question for a moment and get on to the why, the mechanism. Now, to give credit to where it's due, the mechanism was first proposed by one of my best friends, a software engineer named Dave Feldman. And the model is called the lipid energy model. As simply as I can put it, if this is new to you, lipid energy model works like this. When you go low carb, low enough carb to get into ketosis, and you're lean, your body flips rapidly from burning sugar and carbs to burning fat. But fat needs a delivery truck, and that truck includes cholesterol. The more fat you burn, the more trucks you need. And that's why in people like me, LDL skyrockets. Now, if that's enough of a mechanistic explanation for you and you want to get on to the next chapter, you can skip in the video to the section on the Oreo study. But for those who want just a wee bit more nerdiness and technicality, here's a more detailed explanation. Again, I'll try to be clear and concise, but I want to delve a little bit deeper. So when you're lean and low carb, your body very quickly and efficiently swaps from carb burning to fat burning. This change in hormones, or the change in hormones, that occurs with fasting or carbohydrate restriction is actually very different in insulin-sensitive lean people as compared to people with overweight or obesity. And ultimately, this leads to an upsurge in the release of free fatty acids into the blood. Your fat cells pop off the free fatty acids into the blood, and these can feed muscles and be an energy source for, you know, cardiac muscle and also skeletal muscle, but many of these free fatty acids get recirculated through the body via a system that depends on cholesterol-containing particles, including LDL and ApoB-containing particles. So specifically, the free fatty acids get taken up by the liver, packaged into their storage form. These are called triglycerides. Basically, you pin 3 fatty acids onto a glycerol backbone. It kind of looks like an M. And then these, because they don't dissolve in water, get packaged into the core of this big fluffy particle. Called the very low density lipoprotein, or VLDL. And these VLDL particles, they get shipped out of the liver and go into the bloodstream, where they can drop off that triglyceride, that fat cargo, back at fat cells, little fat cells, to replenish them, and at muscle cells to fuel them. But importantly, here's the nuance. In dropping off their fat cargo, these VLDL shrink, they lose the V, and they become LDL particles. And the cholesterol fraction of these LDL particles is your LDL cholesterol. And at the same time, the triglycerides are sucked out of the core of the particle, and surface components, as you're shrinking the sphere, are shed off, including cholesterol that's taken up by HDL, leading to an increase in HDL cholesterol. And the ultimate result is a metabolic fingerprint, a triad of markers: very high LDL, and very high HDL and low triglycerides. This triad is actually the definition of what has been termed lean mass hyperresponder. There's actually no BMI or leanness criteria for a lean mass hyperresponder, but because the phenomenon is so specific to this triad of markers, it's almost a hypothesis. If you see these triad of markers occur collectively, you could actually guess— it's kind of like a party trick— that a person is lean and generally athletic. That's pretty interesting, right? But now this finally brings us back to Oreos, because if you recall, the thing that kicked off this whole flywheel was carbohydrate restriction. When you go low carb, the carb stores, the glycogen in your liver, drops, and this is the stimulus to switch from carb burning to fat burning, kicking off the lipid energy waddle flywheel. So if this is the mechanism, then putting carbs back into the liver should be enough, should be sufficient to slow down the lipid energy model flywheel and as a result drop LDL cholesterol, drop my LDL cholesterol. So that brings us to the Oreo study. And in transparency, in fairness, people had done similar studies before me, my friend Dave, doctors with their own patients using things like fruits and starches, but nobody had done it the way I did it. So I decided to try it with the most salacious, viral, click-worthy food I could imagine, and I came up with, you know it, Oreo cookies. Because we all know Oreo cookies. We all know they're not a health food. They're delicious, at least for some, but they're not a health food. So if I can take a canonical unhealthy food and use it to massively I could tank my LDL cholesterol not as a dietary swap, but as a pure addition, effectively overeating cookies to drop cholesterol. Well, that would be a banger headline and get our research, our broader research, some attention. And let's be clear, to tack on to that, I was a Harvard med student at the time. So you bring Harvard brand PhD, keto, Oreo cookies, and cholesterol together, You got a pretty viral soup, let's be honest. So yeah, my intention was clickbait, but clickbait to invite a conversation about true fascinating physiology, bona fide interesting biology. Now let's get to the actual experiment. What did I do? Well, the experiment was as follows: I locked into a baseline ketogenic diet very strictly for about 2 weeks. Then I went into the Oreo phase, eating 12 Oreo cookies per day. About 100 grams of added carbs. This was a pure addition to my ketogenic diet. I did not decrease my fat intake. In fact, in adding the Oreos, I also technically added some fat, and my saturated fat intake actually went up. Then after this period, I did a washout period where I metabolically reset before the final phase: high-dose powerful statin therapy, 20 milligrams of Crestor for 6 weeks. And throughout, I had my lipids checked. Actually, my primary care physician ordered them directly into my electronic medical record under the oversight of a consulting lipidologist. And what we found was shocking. Not only did the cookies drop my cholesterol, but they dropped my LDL cholesterol twice as much as the statin and in one-third the time. So did it work? It worked unbelievably well. But now comes the tricky part. Interpretation. What do you think this means? That was a gap I intentionally left for readers and the audience to grapple with because I knew this experiment would provoke a cognitive dissonance. Oreo cookies are unhealthy, right? But dropping LDL should be healthy? It should be good? So how can you have a bad unhealthy intervention lead to a good outcome? It doesn't make sense. It provokes cognitive dissonance. It's a paradox. And most people are uncomfortable with that sort of cognitive dissonance. So what do they do? They start to do what is human, to layer on their own interpretations, their own biases. And I knew this would fuel discussion. I was provoking it. I knew it would provide fodder for also confirmation bias and twisted narratives. I'm not naive, but call that necessary collateral damage. I didn't think anybody would actually be convinced Oreo cookies are a true health food. And my ultimate hope was that this engagement would shine a spotlight on an interesting area of broader research. And it did. It worked wonders. It brought us more resources, more partnerships. It got clinicians informed about the lipid-energy model so they can personalize treatments to their patients. No, not cookies, but patients who say we're statin and PCSK9 intolerant could go under sweet potato therapy and drop their LDL by like 480 mg/dL. So this information provided people with power and it stimulated discussion to further research. But now let's take a little bit of a tangent and talk about risk. Truth be told, I still don't have an answer for you with respect to whether or not eating Oreo cookies improved my heart health or not. What I can say is that since that time, our operations have scaled, as I just mentioned. We've had multiple new papers and new initiatives underway. The lipid energy model which I described to you appears to be pretty robust, and I'll link more information on it below, completely open access if you want to read the original mechanisms paper. And with respect to risk, now we've studied more people like me. And while lean mass hyperresponders are a heterogeneous risk group, not everybody is at the same level of risk, the early evidence suggests we might not be at as high cardiovascular risk as one might otherwise think. We're definitely not like familial hypercholesterolemia. I can tell you my own arteries, my own blood vessels in my heart are perfectly clear on coronary CT angiography, no detectable plaque. And I have family members who are also lean and lean mass hyperresponders and much older than me, including my mom who's in her 60s with an LDL around 400, and her vessels are perfectly clear too. And our recent research suggests that plaque progression— this is important— cardiovascular plaque progression is modest in people like me at a population level. And while some people do progress, they do, others have no progression, and some even have cardiovascular plaque regression. And importantly, LDL and applebee exposure are not predictors of how much someone will progress. That's a very interesting, important, and provocative finding. So backing up, What have we learned? What do we know? Well, we do know that Oreos work to lower cholesterol in lean mass hyperresponders. While I've not conducted a formal larger study nor coerced people into replicating these findings, I'll disclose that people have, of their own accord, done so, including with Oreo cookies, and that includes other MDs actually at Harvard. You'll find my peers are actually a very curious group. #staycurious. Anyway, we know that people like me, lean mass hyperresponders, we're unique. We're a very unique population in that we have very high LDL levels and ApoB, but in the absence generally of other risk factors. This is a phenomenon completely distinct from one, the general population who is generally metabolically unhealthy, but also distinct from familial hypercholesterolemia, which is the only other known condition where LDL levels get this high. But that occurs for a very different reason of broken lipid metabolism and genetics, which can also lead to different outcomes. And so what we have here, and this is really the point of this video in the Oreo versus Statin study, is that in people like me, we have an unprecedented opportunity to study the isolated consequences of one of the most vilified biomarkers in history: LDL cholesterol. And let me be clear, this has never been done before, and it has enormous implications for how we think about human cardiovascular health and basic human physiology, because if even one person, let alone a population, can live with LDL cholesterol levels in the 500s and 600s— my total cholesterol's near 700— and live healthily for a long time, it means high LDL alone isn't sufficient to drive cardiovascular disease. It means there are massive gaps in the current model Gaps we need to plug collectively with new knowledge to best serve the health of the entire population. So let me return to the purpose of my experiment. Yes, I engineered this study for engagement. I knew the headlines, "Oreo cookies beat statins," would make people click. But that was the point. Because in a world where attention drives discovery, sometimes you have to use the system to challenge the system. The purpose of the Oreo versus Statin study was to leverage the communication platforms and incentive structures of the modern era to force a spotlight where I thought the spotlight should be aimed. We live in a biomedical world whose spotlight is often directed by money and financial incentive structures. That is the system we have built. But to tap into another way, which I think is possible, I needed something, I need something grassroots. And it comes down to you in one word. And I know you know what that word is. Are you ready to say it with me? Say it out loud in your living room, car, wherever you are. Curiosity. Curiosity. So stay curious. Ask questions, push back. Let's have fruitful discussions so that I don't need to eat another freaking Oreo cookie because I'm going to be honest, and this might be the most controversial thing I say. They kind of suck.