A reasonable conversation with Dave Feldman...

I am writing a response to Dave Feldman’s thoughtful tweet and a post on his YouTube channel. And since I do not have a YouTube channel, I will respond in writing.

First, I want to echo what Dave said about the civility of the conversation and want to thank him for that. I am sure he gets a lot of condescension from doctors and scientists, but I am a big believer in the value newcomers can bring to a field. Some of the best scientific questions I have asked are those that I would not have asked had I “known” better. So, thank you Dave for asking hard questions and for challenging dogma. It is how science happens. 

I will admit that I have followed the story of Dave Feldman and of others for a while. And for a long time I struggled to understand the root of the issue. But I finally do think I now understand. That is probably a good place to start. Dave is a smart engineer and he tells the story of how he decided to try a low-carbohydrate high fat (LCHF) or ketogenic diet 4 years ago. Like many others, he loved the results, as he lost a lot of weight, his metabolic markers improved, he felt more energy and on and on. But, there was a but. While almost all of his cardiometabolic risk markers improved, there was one that did not: his low-density lipoprotein (LDL) cholesterol increased dramatically. I won’t litigate the history of LDL cholesterol here but for those of you who are new to this discussion, LDL cholesterol is among the most studied risk factors for coronary artery disease and if you ask almost any cardiologist or internist, they will tell you that having high LDL is a risk factor for developing coronary artery disease and heart attacks, and low LDL is protective against risk of heart attacks. 

I’ll interject here that with an example where we as a medical and scientific community got it wrong when it came to lipids and biomarkers. This is the story of high-density lipoprotein (HDL) cholesterol and triglycerides. When I was a medical student in the 1990’s we were taught that LDL was bad cholesterol, HDL was good cholesterol, and you could mostly ignore triglycerides. We worked to lower the LDL cholesterol and to raise HDL cholesterol all while paying no attention to triglycerides. Now it appears that was wrong. While the debate about LDL continues, the HDL story evolved over the past few years as one after another drugs designed to increase HDL levels failed to reduce risk of heart attack and even sometimes increased the risk of dying. And then human genetics work from Sek Kathiresan and others showed that HDL was itself not a driver of heart attack risk but was instead a marker of risk. It was guilt by association. And triglycerides, the poor step-child whom we were taught to ignore was found to have direct and causal effects on increasing heart attack risk.

This story is important for 2 reasons: 1) it is an example where the dogma was wrong and 2) this is an example where the established medical community has now admitted they were wrong. Sometimes we hear of vast conspiracies involving doctors and drug companies working to make the world take statins, but the HDL story is a great example where the financial incentives of the large pharmaceutical companies were very much aligned with what was thought to be settled science about HDL cholesterol – science that had been based entirely on epidemiology and basic science or mechanistic plausibility. Yet, the results of the trials matched up very well with the human genetics and the combination has resulted in the complete refutation of what was a long-held belief about the protective role of HDL cholesterol. The companies had to terminate billion dollar drug development programs, textbooks are being re-written and scientific careers are “evolving”.

This brings us back to Dave and his great results with LCHF balanced against a troubling increase in LDL cholesterol. And it turns out that Dave was not alone. Many people were having the same experience and were left with what can best be described as cognitive dissonance balancing the many positive things afforded by LCHF against the potential that it might be causing long term harm. And thus, Dave set out on a journey to learn. But I would suggest that Dave and others really set out on a journey to resolve the cognitive dissonance. I believe that the goal was to make themselves feel better about a diet and lifestyle they loved. There is nothing wrong with that, but we should acknowledge that the goal was to convince themselves that the changes in LDL cholesterol they observed while on LCHF were safe, maybe even protective.

Now is a good place for me to remind folks that I am myself no stranger to LCHF. Being that I am a second-generation cardiologist, and that I grew up in the low-fat era of the 1970’s and 80’s, it was not obvious that I would arrive at the place where I would consider and even adopt a LCHF diet. Yet, I learned about LCHF and saw the many benefits people were having. And for 2 years, I have served as a scientific advisor to Virta Health, and last fall, I co-founded a company called Keyto Inc., a tech-enabled consumer weight loss company based on enabling the use of the ketogenic diet. I have also been on a LCHF diet myself for over a year, and I’ve had great success. I now often recommend it to many of my friends, relatives and even patients. I am a buyer. I am intellectually and financially motivated to see the LCHF diet succeed. I do not want people to stop doing LCHF, and I don’t want there to be risk associated with LCHF. But I am not ready to let go of LDL even though it is dissonant.

One thing about which we can all agree is that there is no gold-standard level evidence that eating LCHF has an impact – positive or negative – on hard cardiovascular disease risk. To have such evidence would require randomized controlled trials that have not yet been done and may never be done. So how do we resolve the critical questions in the meantime? 

One approach is to do what Dave and others have done, which is to resolve the cognitive dissonance over the benefits of LCHF against the potential risk of high LDL. That approach involves going out on a hunt for evidence that will disprove the lipid hypothesis. If Dave can demonstrate that there is not an increased risk (or there may be reduced risk) in some individuals with high LDL, high HDL and low triglycerides, it will make it easier to argue that it is safe to ignore the changes in these lipid markers if you are one of the unlucky few who have this pattern of response. This approach requiresat least partial dissociation of the relationship between LDL levels and cardiovascular disease risk. There are ample examples to point to where that has happened including end-stage liver disease, acute illness, and malnutrition or an intervention such as hormone replacement therapy which improves the lipid markers without a concomitant reduction in risk.

Ultimately this is a conversation about risk. And given that the overall risk of cardiovascular disease is low, especially in younger people, the impact of high LDL might not be manifest for decades. So effectively we are all making a bet. On the one hand, Dave and others are exploring the idea that maybe there is something special about the changes in LDL in in people doing LCHF? Maybe the changes in other markers like triglycerides or insulin or glucose or the improvement in body composition counterbalance the changes in LDL? Or maybe there is something special about lean people who have this very significant increase in LDL on LCHF? In the end, they are discounting the body of evidence supporting high LDL cholesterol as a risk factor for coronary artery disease. 

There are some who take a more measured approach and suggest that the high LDL response can be monitored using a test like a coronary artery calcium scan, a CT angiogram or an ultrasound to measure carotid intimal hyperplasia. And while this is a reasonable consideration, it is very important to remember: 1) the current literature on LDL suggests that it is the lifetime exposure to LDL that drives risk and 2) the use of coronary artery calcium scanning in younger people is not as useful. That is, a calcium score of zero means much less in a 40-year old than it would in a 70-year old. 

This brings me to my approach. And again, I very much see this as a discussion about how to frame and best hedge risk. As of today, I very much believe that the vast body of evidence spanning epidemiology, biology, animal models, gain and loss of function human genetics (polygenic and monogenic) and randomized controlled drug trials supports the role of LDL cholesterol (formally ApoB containing lipoproteins) as being an important and directly causal risk factor for coronary artery disease and heart attack. Do I allow that someday, I could be wrong? Of course, but this is a disease with a very long lead-time, and I have to make decisions now that will impact the future me. And because by the time I learn I’m wrong, it might be too late, I am forced to make a decision based on the information I have today, not the information I wish I had. And given the information I have here today, I have decided that I do not want to dismiss LDL as a risk factor for heart attack. 

Does this mean I need to stop LCHF or we should shut down Keyto? No! Of course not. Luckily, only a small subset of people who go on a LCHF diet end up with a significant increase in LDL. And luckily there are other options. This point is critical. This is not a binary decision: either die from a heart attack or seek to disprove the lipid hypothesis.  If people like the way they feel or look or they like the changes in other risk markers, they certainly to not have to stop. Again, there are other options. 

The first option is probably unpopular, but it would be to take a statin or other lipid modifying medication. Statins are red-hot controversial now for all sorts of reasons, but let’s review facts. The first statin was approved in the late 1980’s. Hundreds of thousands of people have been randomized to clinical trials examining the effects of statins on cardiovascular and other diseases. Some of those trials have been negative. But overall, there is roughly a 20% reduction in risk of heart attack in people taking statin for every 40 mg/dl reduction in LDL cholesterol. Should someone with a 1% 10-year risk of heart attack be taking a statin? No. It is hard to reduce a 1% risk to less than 1%. But we have also learned that, when tolerated (this is an important caveat), statins are extremely safe. In fact, there is no strong safety signal of any significant adverse outcome (other than some very rare ones) aside from a very modest increase in the risk of diabetes which appears to be related to the LDL reduction as it is seen across all LDL-lowering drug classes as well as in people with genetically low LDL

What about tolerability? This is a critical question and it is one that will hopefully be answered in this very creative trial. And I will not at all dispute that some people do not feel good when they take a statin. But that is why I frame all conversations with patients about drugs as being an experiment, not a contract. If you try it and feel bad, stop. If you don’t feel anything, you can always stop later. If there is new evidence that a drug has a risk, you stop. You are not signing a contract to take this drug for the rest of your life. If you take a statin and it has no adverse effect that you feel or that we can measure, why not take it? What is the harm? This decision is a hedge: The risk of high LDL over decades is or is not harmful versusstatin that you cannot feel and has no measurable side effect is or is not helpful.

But I also understand that many people do not want to take statins or any medications and many will not need to. So luckily there is another option. Much research suggests that the observed LDL increase in people on LCHF diets is the specific result of saturated fats. And some have found that reducing the ratio of saturated fats to total fats by replacing some or most of the saturated fats you eat with mono- and polyunsaturated (especially omega 3) fats mitigates the LDL increase, at least to some degree. 

In the end, we need more research and I hope to be a part of doing it. But until then, we are left having to make important decisions that may impact our risk of having disease 10, 20, 30 years from now. We have to make decisions with the evidence available today and not evidence we want or wish we had. And what I would hope is that people use the best available information to make the best-informed decisions for them. I also hope is that it is not necessary to completely undermine the lipid hypothesis in an effort to solve a confusing and complicated conundrum that applies to what is today a very small group of people. I am working on a piece with Nicola Guess on this very topic that I hope we can publish soon. But until then, let’s continue to have this conversation, and let’s continue to ask hard and good questions. Thanks Dave for engaging with me in a very thoughtful and respectful manner, and I look forward to the next chapter...


  1. Done all the same n=1 research you mention. Statin intolerance drove me to paleo in 2008. Added guar gum, psylium to reduce LDLs from 260 to 230. In 2012-2017, LAD scores started to increase from 0 to up 30% a year on LCHF diet, so I scaled back meat 90% and replaced with sprouted grains and sweet potatoes. LDLs dropped 40% in 2 months to lowest ever for me as a familial hypercholesterlemic individual. Triglyerides only moved up from 50 to 68 while LDLs dropped from 170 to 110 (on highest dose pcsk9 + guar gum, flax husk, and psyllium).

    1. These are impressive results. After doing keto for 15 months, I too started to reintroduce some slow carbs--including sweet potaroes, sprouted grains and beans. While I was no longer in ketosis, this change in my diet helped me to lose an additional 25 pounds after my weight loss stalled for several months on keto. It also helped with food boredom by expanding my food choices. I am still a huge keto advocate, and plan to return to it this summer to lose that final 20 pounds, but I think its good to make individual tweaks to enjoy the full benefits of LCHF.

  2. It often strikes me how the keto diet is construed as being a single diet, when I really think it is a spectrum of diets. BlueStar's comments above illustrate this very well. So many people in the keto community hold fast to the idea that the keto diet that most people practice--lots of fatty meats, bacon, butter, etc--is sacrosanct, and ideal for everybody. Sure, you can knock down the significance of a high LDL, as being ohhhh-soooo-1990, but I would rather hedge my bets by adjusting my diet to address the problem. Others simply say the problem isn't really a problem. I got fat on a junk-food vegetarian diet, and I got slim on a keto vegetarian diet. My lipid profile was pristine throughout my life, but I did see my triglycerides drop from 119 to 61, and my HDL went from 69 to 61, total cholesterol remained at 140, or so.

    1. Typo there. HDL went from 61 to 91. There, that's better!

  3. Keto diets will be a good long term fit for a minority as it restricts half the foods usually consumed. The time tested dietary pattern enjoyed by the healthiest populations in the world remains the Mediterranean style diets and this diet is consistently ranked easiest to maintain. Worth recommending patients start here.

  4. I enjoyed reading this. I think it's a mark of a rational mind to be able to consider that something that benefits you may have a downside, and be willing to manage that downside. I've observed that most people are very resistant to the idea that their lifestyle choices might have downsides.


Post a Comment

Popular Posts