
Last year at Low Carb Gold Coast, I had the pleasure of watching an excellent presentation by Dr. Paul Mason where he talked us through hyper-responders and lipoproteins. Naturally, I was loving every second. To my surprise, he addressed the Feldman Protocol as well, with a suggestion of adding carbs to achieve the same outcome.
Here’s a transcript of that portion of the talk (at about 21:54):
Now we’ve all heard about the Feldman protocol.
So essentially what happens is if you have a very high LDL level on a ketogenic diet, it’s been demonstrated, quite nicely, that if you go on a very high fat diet for about three days and have a blood test at the end of it your LDL levels will significantly drop, some would even say “plummet”. And this is why – it comes down to these LDL receptors.
Now these LDL receptors are what actually takes healthy LDL out of the bloodstream it doesn’t work for damaged LDL, remember, it only works for the healthy LDL and the amount of LDL in the blood has an inverse relationship to the number of receptors. If we have more receptors that are able to take the LDL out of the circulation then we’re going to have less in our circulation. It’s only logical. And interestingly, increasing the amount of calories in our diet, which increases our insulin, actually increases the genetic expression of these LDL receptors.
In actual fact, rather than doing a high-fat diet as with the Feldman protocol a lesser amount of carbohydrates would probably do exactly the same thing because of much stronger insulin response.

Now, why does it take three days? Well just because you increase the expression of a gene doesn’t mean that you get an instant effect there’s a lot of steps that you need to go through before you end up with the final product, which in this case is the LDL receptors and it’s this process between genetic upregulation and the final protein synthesis that likely explains the three-day delay between when you increase the amount of energy in the diet and the LDL actually falls.
Paul’s theory is pretty sound. Again, we’re reading a lot of the same literature. But that said, I had already done an experiment that added carbs to a keto diet known as the Added Sugar Experiment. The net effect wasn’t a significant drop in cholesterol.
However, in chatting with Paul since his talk, we theorized there might be a component with the fact the skittles were far more fructose-based and this could be a confounder. So we discussed using something with glucose, dextrose, or maltose.

Bottlecaps to the rescue! Fortunately, many Willy Wonky candies have dextrose as their primary sweetener, so we agreed on using this brand as the intervention.
Study Design
I was just coming off a carnivore diet (more on that in an upcoming presentation and write up). So I worked out being on my “baseline” diet for five days, then following it with 4 days of the baseline with the addition of the Bottlecaps.
[Paul and I discussed this in an ad hoc video before performing the experiment which you can watch here]



Changes in Glucose on Dexcom G6 (CGM)
The value of continuous glucose monitors (CGM) is that they can give a constant reading on what is happening to your glucose levels at five minute intervals. I was well aware this would likely jump after I added the dextrose.
For reference, here was a typical 24h reading during my baseline diet, before adding the sugar:

Alas, Dexcom offers no y axis in this view, but I can tell you that my line basically hovers between 85 mg/dL and 105 at its peak. Pretty much the flattest line I ever make on the CGM. (Hello, keto!)
Then Day 1 of the added candy hits and I’m running a little late, hitting the first meal at about 11:45. I get this response following the first meal:

Whoa!
You might recall in my end of the Tandem Drop Experiment I was pretty uncomfortable with topping out at 183 mg/dL on my CGM. Nevermind… that was just a warm up…
This is the full 24 hours as shown before I started my meal on the next day:

I topped out at 255 (last arc) and woke up to a 108 the next morning.
As I told Paul, I seriously considered canceling the experiment at least three times on Day 1. I was certainly uncomfortable with these numbers and that can be a confounder in and of itself (such as raising cortisol, et). But ultimately, I decided to move forward.
Side note: I originally thought I topped out at 254 for Day 1, which led me to post this poll on Twitter:
After 4 weeks of ultra low carb #carnivore, I began a 4 day experiment yesterday involving 75g of dextrose with each meal, 3 times a day. What do you predict my glucose topped out at during the day yesterday? (Results revealed in write up)
— Dave Feldman (@DaveKeto) January 15, 2019
Thus, 76% of people (myself included) would’ve picked an answer below 254.

By Day 4 I had seen a smaller top off at 232 mg/dL, but I’d call that only a marginal improvement.
It is common advice to those on a low carb diet to consider eating lots of carbs in the days leading up to an Oral Glucose Tolerance Test (OGTT) as it will acclimate one to having a less dramatic response with glucose. Given my own results from above, I’m not sure if this would have worked out so well for me in the context of added carbs. But perhaps it would’ve with swapping in carbs for swapping out fat.
Cholesterol Changes
Now on to the Main Event – LDL Cholesterol:

We did indeed see a drop, although not a very dramatic one when compared to baseline. The last day of the baseline came to 313 with the added dextrose at day four clocking in at 265 for a difference of 48 mg/dL or a 15.3% drop.
Our top three comparisons are my first Feldman Protocol (drop of 73 / 28.5%), the Energy Status Experiment (drop of 73 / 30.1%), and Tandem Drop Experiment (first three days – drop of 136 mg/dL / 45.9%).

The Weight Confounder
Alas, there was also another possible reason for the drop in LDL-C. After I added in the Dextrose, I began gaining weight… pretty rapidly, in fact.
Not too coincidentally, I ran into this very phenomenon in reverse with my Weight Gain Experiment from last year. While I was actively losing weight, I saw a tight inverse correlation with my LDL-C rising (Pearson of -0.936!)
Naturally, at higher levels of insulin we have greater inhibition of lipolysis, leading to higher weight retention. As this lowers the free fatty acid pool, we likewise have less VLDL production and thus less downstream LDL. In theory, this could account for some or quite a lot of the lower overall resulting LDL.

Fortunately, I got a Free Fatty Acid test with all my bloodwork, so we’ll have that for a comparison once all the labs are back, which will likely be in a week or two.
Symptoms
There were a number of issues with this experiment that were notable:
- Initial anxiety. This was likely due to my reacting to the CGM readings, which subsided after the first day. But worth noting nonetheless.
- Mild clumsiness. This was something very unexpected, yet distinctive enough to journal. With each day past the first, I had one or more moments of stumbling or letting things slip out of my hand in a very uncharacteristic fashion.
- “Shipping” issues on Day 2. (Yes, I’m using a euphemism here for bathroom problems.) I was quite surprised it happened so quickly, just one day into the intervention. Because of this, I did have to consume magnesium citrate, which did resolve the issue and it didn’t return again for the duration of the experiment.
Would a Longer Time Make a Difference?
Paul wanted to emphasize that he’d prefer I did the experiment for seven days instead of four. And indeed, since my previous interventions three days, it’s possible he could be right that a more dramatic change was up ahead. But frankly, I just didn’t have it in me. It was tough enough to finish out these four days watching my CGM going off the charts. But more problematic would be if I continued to gain weight, which would have been a persistent confounder.
Moreover, if indeed a longer time is needed, this would seem to suggest any of the other protocols above would be more attractive given turnaround time of just a few days instead.
Final Thoughts
- Once again, we seem to have further evidence that swapping is preferable to addition with my N=1. More specifically, if you’re going to attempt using carbs to lower your LDL, you might be much better off swapping out the calories from fat to make room for the calories from carbs. How much of this relates back to weight gain may need further research to determine.
- It’s possible this experiment reflects issues with carb loading in the days before an OGTT, but there’s a catch. At first glance, it appears that this process didn’t help me acclimate to a higher carb diet. Yet these spikes were always in the context of the dextrose load following a fatty meal. And, of course, it wasn’t consumed all at once, such as in a sugary liquid bolus. Any of these may have contributed to a very different outcome.
[Paul and I discussed the results in a video at the conclusion of the experiment which you can watch here]
I really want to think Paul for working with me on this. While it wasn’t as dramatic as we had hoped, it is worth reemphasizing that this did indeed succeed at lowering my LDL as hypothesized. Further, it suggests there was a difference between the fructose-based sweetener from Skittles (in the original Added Sugar Experiment) vs dextrose-based sweetener from Bottlecaps.
So perhaps Willy Wonka can thank us for giving them a new tagline on how their candy lowers cholesterol, unlike the competitors.
I had to work out what was happening to my blood glucose longhand using a glucometer as CGMs weren’t around 14 years ago. I soon found my BG would shoot up after eating any substantial quantity of carbs – it would be at its highest about 1 hour after eating, be substantially back to normal at 2 hours, when diabetics are told to test, then at three to four hours it would drop significantly below normal, and the worst symptoms I got were from the speed of the drop not necessarily from the low per se. Thanks to sensible people, some of whom had sensible doctors, I realised the body will intervene to avoid a hypo before it arrives by dumping “counterregulatory hormones”, not just glucagon but cortisol, epinephrine, norepinephrine and others. This may be what caused your symptoms of anxiety and clumsiness. Kinda like “hangry” on steroids.
I am new to this web site , so this is off topic a little bit and perhaps can just be reposted where most relevant. The original publication is actually from about 10 years ago, but only, now, available at the AHA Circulation web site: “Abstract 4133: Dietary Change Induced Weight Loss Produces a Dramatic Rise in Serum Homocysteine Levels: Modification by Folic Acid and B Vitamin Supplementation” ( SR Gundry ) (https://www.ahajournals.org/doi/10.1161/circ.114.suppl_18.II_892-a)
Then we have a publication from Jan. 2019 advocating homocysteine levels as a marker of diabetic retinopathy: “Homocysteine: A Potential Biomarker for Diabetic Retinopathy”
(https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6352029/)
So let’s connect the dots where Dr. Mason says the tri/HDL ratios correlate to the Kraft abnormal insulin patterns, and if this is so this strongly suggests the need for a dilated eye exam for an individual. Eye clinics can compare the data already available in patient records to see how strong the correlation is to level of diabetic retinopathy, that is the tri/HDL patterns. Point is the data is already there and does not need to be collected.
Also now we have the elevated homocysteine due to rapid diet induced weight loss, and perhaps damage to the retina from the elevated homocysteine.
Pressing need then to get individual methylation snps and make use of P5P, folate, betaine, and B12 to get the homocysteine under control – besides ancestral diet.
I also have the caffeine rs762551 AC snp and no problema at all with tri levels last two years, after 3 years keto. Tri/HDL ratio was .66 last year and .75 this year so paying lots more attention to the homocysteine problem. It dropped two points over last 2 years, but still at 9.6 and I have a ways to go to get it down to 6, like Dr. Bredesen recommends.
Dave I just put together some of this, in my mind, as I was writing and thanks for the attitude of open enquiry , that you encourage so well. Nothing like a team effort to get at the root of a problem.
Hi Dave,
I applaud your determination – effectively 3 OGTT per day – wow!
May I ask – do you have any idea what your INSULIN was doing during those stellar BG excursions??
I can see two possibilities:
1) Your insulin was much elevated and this was the best control your metabolism could achieve…
OR
2) You had delayed first-phase pancreatic insulin response – it took time for your insulin to ‘crank up’ to tackle the BG challenge…
There is an important reason why I ask, but first I am curious about your opinion…
Thanks
Nick
Looks just like my OGTT with delayed insulin response followed by hypoglycemia. DX was delayed reactive hypoglycemia. That was 2+ years ago. Seeing it again now that I’m trialing a Freestyle Libre. Today I’ve been reading that bg excursions are correlated to CIMT values. I had been thinking that a couple of hours a day of high blood glucose followed by rapid lowering was probably not particularly harmful. Definitely rethinking that now.
Grest ifile here…. However, I’m assuming you’re not using a responsive WordPress theme.. My mobile browser renders this site as an old fashion desktop website. This makes it difficult to read. Too much left/right scrolling.
I believe I am a hyper responder. Please let me know if you are still looking for people for your study.
You might want to check out the Lean Mass Hyper-responder facebook group – there is sign up for a potential project, although not many details are out yet.
Considering some of the pretty quick improvements in glucose control that plant-based doctors have gotten in their studies, I think that replacing fat for carbs would’ve worked a lot better. If you’re up to it, it would be interesting if you tried eating low fat (<10%) and do the same experiment.
This is related to substrate priority and the Randle cycle: https://en.wikipedia.org/wiki/Randle_cycle.
He did do a white bread and lean meat experiment! (Carb swap)
Hi Dave,
I just watched your “Added Sugar Experiment 2 Findings – with Dr. Paul Mason” from earlier this year and this “Weight Confounder” thing didn’t seem like it should have been a surprise.
My experience is: I’m 43yrs old, 5’11”, ~250lbs, and been doing Keto for ~3.5yrs. I typically do a weigh-in 1st thing in the morning (which my Fitbit Aria 2 scale dutifully records for me).
My experience has been that whenever I seriously bust through my carb budget, I see a weight gain in the next day or two of ~2-4 lbs that lingers for ~2-4 days with an end often marked by an opposite effect to your “shipping issues”.
My understanding has been that this is due to changes in my glycogen stores (which I’ve heard described as a watery gel for suspending glycogen): building me up with water weight in response to the carbs, lingering for a few days after the initial event while going unused, before then being retired along with all the associated water weight.
Dr. Paul Mason had even suggested that glycogen storage could have noticeable affects during your ad hoc “Added Sugar Experiment 2 – the Paul Mason Edition” video. Wherein he had suggested that up to 500g of glycogen may be stored as a result of the experiment.
This would be in-line with my experiences, implying another possibly ~1000-1500g of water weight for those 500g of glycogen. Much like that ~4 lbs weight gain number seen during your experiment.
Apparently, the body also uses ~18 mg of potassium per gram of glycogen stored, so a rapid change in glycogen storage could eat up/release significant amounts of potassium (which could also mess with the body’s magnesium regulation). Maybe this is part of the mechanism behind the “shipping issues”?
Anyway, with my (possibly flawed) model, I’d hypothesis that:
1. your “Weight Confounder” increase, was just this glycogen + water weight increase, and not an increase in your “fat” or “muscle” mass.
2. you would have seen your weight stabilize as your glycogen storage capacity maxed out, around +4 lbs, had you continued for more days.
3. if you were to somehow compare your body composition before and after the weight gain (assuming you ran the experiment again?) with something like a dexa scan before and after, it would prove this out.
Maybe it’s just me, but it seems there’s a lot more talk (and understanding?) about what changes to expect while transitioning into a fat based metabolism (becoming fat adapted), than about what changes to expect while transitioning back into a carb based metabolism (becoming carb adapted?).
In the same way that in becoming fat adapted, your body gets better at buffering/regulating the fat energy in your blood between meals, shouldn’t it be expected that in becoming carb adapted you’d see your body get better at buffering/regulating the glucose energy in your blood between meals?
If I recall correctly, your recorded blood sugar spikes/dips did soften as the experiment progressed.
Was this because you ate the Bottlecaps over a longer period of time?
Was this because your glycogen stores were getting better at buffering the spikes and dips? (having gained more capacity allowing them to more strongly reduce spikes and having gained some glycogen content that could be released during the dips)
Was this because of some mix of the two and/or maybe something else?
Anyway, I love hearing about your N of 1 experiments. Thanks and keep up the good work!
-S