Sep 17

Mega Monster #FeldmanProtocol Results

One Upmanship

Ever since I saw Dave’s Fasting Disaster post I’ve wanted to add some fasting data to the Cholesterol Code reserves. Instead of fasting for two days as Dave had done, I wanted to fast for the three days he had originally planned for. In fact, why not do a full five days as outlined in the 10 1/2 day Feldman Protocol instructions?

I had two advantages over Dave: 1) I’m much less lean than Dave is, which would likely make fasting easier. 2) I’ve done multiple multi-day fasts, making me a more advanced “faster”. Fasting is often described as a muscle that you need to train, which I’ve “flexed” a bit more than Dave has.

The opportunity to get more data was too good to pass up. Especially since our data guru (and resident Lean Mass Hyper-responder), Craig, had already completed the full protocol not too long ago, leaving me as the last one out. I was plenty ready and willing to fix that.

The Plan

Instead of following in Dave’s footsteps exactly, I made a few tweaks:

  • Electrolytes on an as-needed basis

Several conversations with Megan Ramos at various conferences cemented the idea that when fasting, electrolytes are no joke. As such whenever I was feeling a bit “off” I would supplement electrolytes.

  • Feeding would follow a carnivorous diet

I’ve been following a carnivorous diet since October of 2017, and thus in order to make the results comparable to my previous data sets, I wanted to keep diet relatively the same as well.

  • No liquid fats

Liquid forms of fat can sometimes mess with triglyceride levels (which can influence calculated LDL), so I decided to avoid them entirely. All of my fat would be found in the meat and cheese I was eating – no butter in my coffee, no chugging heavy cream (as I did during the Ketofest experiment), and definitely no MCT or coconut oil.

The Inversion Pattern

In case it wasn’t obvious, I was setting out replicate the inversion pattern via the Feldman Protocol to see if the observed relationships held true. Going off of Dave’s original experiment from 2016, these relationships include:

  • Total Cholesterol tracks with the inverse of dietary fat for the 1-3 days before the blood draw. (87% inverted correlation)
  • LDL-C tracks with the inverse of dietary fat for the 1-3 days before the blood draw. (90% inverted correlation)
  • LDL-P tracks with the inverse of dietary fat for the 3-5 days before the blood draw. (80% correlation)
  • HDL-C tracks with dietary fat for the 1-3 days before the blood draw. (74% correlation)
  • HDL-P tracks with dietary fat for the 3-5 days before the blood draw. (correlation not calculated)
  • TG tracks with the inverse of dietary fat for the 1-3 days before the blood draw. (61% inverted correlation)

Murphy’s Law

What can go wrong, will go wrong… After I had already started the experiment, and right before my first blood test, I discovered that my local LabCorp wasn’t open on weekends. This entirely threw off my intended schedule, as one of the tests for the feasting portion would have landed on a Saturday. Not only that, but because they were only open on weekdays, this meant the 10 1/2 day protocol was essentially impossible. Unfortunately, there was no other local LabCorp nearby, so I was left with two options.

  1. Do the 6-day protocol instead.
  2. Extend my fasting phase to 7 days to push the Saturday test to Monday.

Because the whole point of the experiment was to get fasting data from at least 5 days of fasting, I decided on option two. I decided I likely had enough body fat and experience to get me through 7 days of full-on fasting safely, and easily, and it would introduce a unique opportunity to get even more data.

With that, the new schedule looked like this:

Food Tracking

As per usual, I tracked all of my food from the high-calorie days (and all electrolytes/beverages during fasting days) through picture-taking. At the end of each day, I also logged all the food into My Fitness Pal. I ended up eating much more cheese and processed meat than expected, but still achieved my goal of at least 3000 calories for the 5 feasting days.

An example of items consumed during fasting phase vs high calorie phase

The Results

Total Cholesterol

First up is Total Cholesterol.

  • Normal Baseline: usually around 320-350 mg/dL (not shown in the graph).

After three days of fasting my Total Cholesterol was 345 mg/dL, where it stayed in about the same range until I switched over to the high calorie/high fat phase where it initially dropped to 219 mg/dL after three days, and then 209 mg/dL after 5 days of the high fat protocol.

  • 7 day fasted to 3 day high fat/high calorie: -132 mg/dL
    • Time span of drop: 3 days
  • Biggest drop: -162 mg/dL (Between highest on 8/22, vs the lowest 8/29)
    • Time span of drop: 7 days

The correlation between my 3 day average of dietary fat was an astounding -0.9928~ even higher than the expected 87% inverse correlation.


Next is LDL Cholesterol.

  • Normal Baseline: usually around 270-290 mg/dL (data not shown).

After 3 days of fasting my LDL-C was 278 mg/dL which climbed to 310 mg/dL after 5 days of fasting. After 3 days of high calorie it quickly dropped to 166 mg/dL and after 5 days of high calorie dropped even further to 151 mg/dL.

  • 7 day fasted to 3 day high fat/high calorie: -132 mg/dL
    • Time span of drop: 3 days
  • Biggest drop: -159 mg/dL (between 8/22 and 8/29)
    • Time span of drop: 7 days

The inverse correlation between LDL-C and the 3 day average of dietary fat was, again, higher than expected at -0.988~ with the expectation being an inverse correlation of 90%.


Particle count!

[IMPORTANT REMINDER: again, the Inversion Pattern for LDL-P is usually a three-day window with a two-day gap, not the three-day window with a zero-day gap, hence why the graph below doesn’t start with dietary fat at 0]

Normal Baseline: somewhere around 3000 nmol/L (not shown).

After 3 days I was actually lower than my usual at 2890 nmol/L, however, this quickly skyrocketed to >3500 nmol/L after 5 and 7 days fasted. For those of you who were curious, if getting an NMR Lipoprofile, the test that measures LDL-P among other things, if you go above 3500 nmol/L it won’t give specifics after that. After 3 days of high fat/high calorie feeding it dropped down to 2086 nmol/L and then to 1578 nmol/L after 5 days of high fat/high calorie.

  • 7 day fasted to 3 day high fat/high calorie (estimated): -1414 nmol/L
    • Time span of drop: 3 days
  • Biggest drop (estimated): -1922 nmol/L (between 8/22, 8/24 and 8/29)
    • Time span of drop: 5-7 days

Unfortunately, those numbers are estimated because of the LDL-P cutoff, each estimated “drop” assumes that LDL-P was at exactly 3500.

Again, if we assume LDL-P was exactly 3500 nmol/L the correlation comes out to -0.91~ however, if we assume both topped out LDL-P are >3500 as a general guess, the correlation drops to -0.83~ which pretty closely matches the expected inverse correlation of 80%.

Note: Due to formatting reasons, the below graph assumes LDL-P was exactly 3500 at its highest


  • Normal Baseline: My normal HDL-C has always tended to run low, with my normal hitting around 40-45 mg/dL on average
    • (for those wondering, the best guess for now is it is genetic, but I’m not entirely sold on that as of yet).

As predicted, HDL-C fell during the fasting days, from 43 mg/dL after 3 days of fasting to 37 mg/dL after 7 days of fasting.

  • 7 days fasted to 3 days high fat/high calorie: +4 mg/dL 
    • Time span of increase: 3 days
  • Biggest Increase: +9 mg/dL (8/24 to 8/29).
    • Time span of increase: 5 days

Although this may not appear to be a lot of movement, HDL is one of the less noisy markers and tends to remain a bit more stable over shorter periods of time. The positive correlation to a 3 day average of dietary fat was slightly lower than the expected 70%, but came in close at 0.66~.


Normal Baseline: Like HDL-C my HDL-P tends to run low, around 20-23 umol/L

HDL-P did fall during the fasting phase, from 16.9 umol/L to 13.2 umol/L with a slight bump up at the 5 day mark, and went up during the high fat/high calorie phase to the highest HDL-P I have on record at 29.1 umol/L

  • 7 days fasted to 3 days high fat/high calorie:  +12.7 umol/L
    • Time span of increase: 3 days
  • Biggest Increase: +15.9 umol/L (8/24 to 8/29)
    • Time span of increase: 5 days

The positive correlation to dietary fat (with a 2 day gap) on this one was 0.84~ with no previous correlation on record, although Craig mentioned in his post the correlation was higher than his HDL-C as it is in mine.


Normal Baseline: 70-90 mg/dL

As expected, triglycerides went high during the early portions of the fast and then started to trend down the longer the fast went on. From 119 mg/dL at the 3 day mark, trending down throughout the fasting phase until it reached 94 mg/dL by day 7 of the fast. It continued to drop through the high calorie phase until it reached my lowest triglyceride level on record at 49 mg/dL.

  • 7 days fasted to 3 days high fat/high calorie:  -33 mg/dL
    • Time span of drop: 3 days
  • Biggest drop: 70 mg/dL
    • Time span of drop: 9 days

Although typically triglycerides are much noisier than usual, the correlation here was -0.95~ with 3 day average dietary fat, exceeding the expectation of a 61% inverse correlation.


Normal Baseline: Usually around 112-140 nmol/L although it has been as high as 187 nmol/L, and as low as 82 nmol/L but neither of these occasions were normal baseline readings.

I actually didn’t expect lipoprotein(a) to fluctuate that much, as with my past data the only things that have moved it so far is getting sick (causes an increase) and swapping meat sources (causes a drop). Generally it’s said in the literature that lipoprotein(a) levels are largely determined by genetic factors, although it does act as an acute phase reactant. As such I expected it fluctuate maybe 10 nmol/L as it usually does when I’m not explicitly trying to move it. But, of course, it had to surprise me by increasing above my normal baseline to 189 nmol/L by 5 days of fasting, then decreasing substantially to 77 nmol/L upon high fat re-feeding, and even further to 65 nmol/L after another 2 days – the lowest lipoprotein(a) readings I’ve ever gotten, leaving me technically “in range” of normal levels.

  • 7 days fasted to 3 days high fat/high calorie:  –103 nmol/L
    • Time span of drop: 3 days
  • Biggest drop: -124 nmol/L (8/22 to 8/29)
    • Time span of drop: 7 days

The correlation with lipoprotein(a) and 3 day average of dietary fat was a pretty impressive -0.998~. There’s no previous correlation on record for Craig or Dave, so as far as I know, this hasn’t been replicated yet.

It has been suggested by some who saw the result that fasting beforehand could have confounded the high fat feeding data. In addition to that possibility, I haven’t replicated the data myself yet, so I’ll definitely have to do a few more experiments to see if this reaction is consistent. For now, though, it certainly is unexpected – not to mention interesting – and it makes me glad I food/calorie matched for the blood draws I got late last year and early this year, while sick.

Thyroid Changes

During the wide spectrum testing days, I decided to check out a few additional markers beyond the “basics”. This included some thyroid markers, to see how they would change before and after the fast. I expected that T3 would be low, as lower T3 might be useful for muscle sparing and energy regulation, but I wasn’t sure what the other markers would look like.


Thyroid Markers 7 day fast 5 day feast Ref Range
TSH 3.36 2.51 .450-4.5
Reverse T3 34.2 14.5 9.2 – 24.1
Thyroxine (T4) 7.1 6.1 4.5 – 12
Triiodothyronine (T3), free 1.6 2.2 2 – 4.4

As expected, T3 went from so low it was out of range, to the lower end of normal (which still didn’t surprise me considering I’ve seen discussion that lower T3 on a ketogenic diet could be adaptive). Reverse T3 dropped by over half, and upon a little poking around it seems this is not unusual and may go hand in hand with the lower T3 as an adaptive change.


In short, I would say that after full completion of the extended protocol, this experiment worked as a confirmation of the protocol’s effects in 1) a woman 2) who has a slightly higher estimated body fat percentage than Craig or Dave 3) who follows a predominantly carnivorous diet. Via the protocol, I successfully dropped my LDL-C by 159 mg/dL in a much shorter time period than would be conventionally assumed plausible. Additionally, I did this via diet changes only, with no changes to supplements (vitamin D, and magnesium glycinate) with the exception of an electrolyte supplement which was consumed on an ad libitum basis. Nor were there any changes to medications (none) during the experiment.

Additionally, I found that the lipid system was far more dynamic for me personally than I had expected. The previous time I had done the protocol I only dropped my cholesterol by about 50 mg/dL (6 day protocol with low-calorie instead of fasting), and I had thought that this attempt would yield a slightly higher drop. In fact, somewhat surprisingly, I nearly tripled the drop in cholesterol this time around.  Additionally, unexpected lipid markers (such as lipoprotein(a)) showed a surprising amount of – what could turn out to be – dynamic response to diet as well. This will obviously require further follow-up to confirm it was the introduction of high amounts of dietary fat that resulted in personally historically low lipoprotein(a) but the initial results of this experiment are intriguing, to say the least.

Leave a Reply

Photo and Image Files
Audio and Video Files
Other File Types
10 Comment threads
14 Thread replies
Most reacted comment
Hottest comment thread
9 Comment authors
Siobhan HugginsAngieAngie AbbacchiPhilippaAdriana Recent comment authors

newest oldest most voted
Notify of

This really makes me question the validity of the cholesterol tests. How does one know they’re getting good data and seeing meaningful results from their nutrition? I regularly do 1-3 day fasts and time restricted feeding days. What’s the best way to get stable, actionable information from the NMR if following keto and fasting? My conclusion after reading some of the other posts on this was that one shouldn’t fast right before the test. Now I’m thinking maybe don’t fast 5-7 days prior is better?

Wow, a 7 day fast is impressive! The most I’ve been able to achieve is 5 days. I usually start to feel cold on the second day and jittery at about day 3. I supplement with 400 mg magnesium glycinate, 198 mg potassium, and 2-4 g sea salt. Playing with those a bit doesn’t seem to help. I have plenty of extra energy stored, so I’m not a LM faster.

I love the freedom and energy (when in the sweet spot) of fasting and want to get to a point that I can easily do five days at a time. Is there a post somewhere that describes your fasting protocol in detail, including supplements?


Wow you go girl!!! Well done.


Hi Siobhan, I also have high Lp(a). I just started testing it this year and it was in the 200-300 range from May through July. I’m particularly concerned because both of my grandfathers died of heart disease. One was a dairy farmer and had heart problems all of this life and the other one smoked most of his life. My sister had her Lp(a) checked after I told her how high mine was and hers (like your sister) was much lower (30ish). All of my other markers are good (trig/HDL =<1, insulin<5, A1C < 5.3, low CRP, low PLAC). I also had a CAC score of 45. Interestingly, my ferritin is a little high (155) which according to your results seem to track with your LP(a). I've thought that red meat might be an issue but I don't have any homozygous genes for iron overload. Anyway, I'm getting some blood work done this week to see where I'm at now. I'm not keto but try to lean that way when I can. I'll let you know if anything changes and if I can move LP(a) as much as you have.


Hi Siobhan,

Well I wished I had results like your!! Just got my results back and I suspect coffee!!
Female 55 on keto for more than a year. I am definitely a hyper responder…
My LDL went from 5.42 to 5.70 my trig from 1.01 to 0.75 and my HDL from 1.65 to 1.96. Even my keto doc is freaking out over my high LDL… I am switching to carnivore and see who that goes… Will calculate my remnant.

Keep up your good work.


oops forgot to say that before LCHF and Keto my LDL was 2.80 my HDL was 1.71 and my trig were 0.95. But ironically I just calculated my remnants and it’s gotten better at .34 mmol/L in the lowest risk Quintile
Remnant cool to HDL is 0.17 and AIP at -0.417. Oh I sure hope you guys are right about this.


I hope you don’t mind questions like this….maybe someone else’s question too. I want to do the Feldman protocol. I have orders from my dr for routine blood work and asked for extra thyroid tests (while insurance was paying). I want to do the high calorie part of the test so the doctor will see the lower LDL… My questions is, will that mess up my thyroid results? Are thyroid numbers as fluid as cholesterol?

Amy Blackwell
Amy Blackwell

This is really great work! Thanks so much for exploring this area and writing up your results in such details. It’s a very valuable contribution in a field that has been ossified with imaginary information for decades. It’s fascinating to see just how dynamic the lipid system can be.

The Tandem Experiment Part 1: Fat » Cholesterol Code

[…] of starting out with fasting, as I did with my Feldman Protocol experiment I instead decided to start with Ketochow intake with baseline calories in the 4 days […]


Just dipping my toe into Feldman protocol info here. How many callries and grams of fat were you consuming daily? During the “2-day gap” did you continue to eat high fat/high calorie? How important do you think the fast was to your results? I wish you had blood work done before the fast as an initial baseline to see if the fasting made any difference.


Hi Siobhan, Great data thanks for fasting for science!
I was wondering if you would mind sharing the CBC data for 5 days fasted (particularly the absolute lymphocyte count) The youtube video skips over the 5 day CBC.. Also do you have a baseline lymphocyte count from your last normal blood draw before the experiment (or an average of your normal baseline)?
The reason I am interested is that Valter Longo has seen a drop in lymphocyte count with fasting (or FMD) that reverses as the immune system regenerates after refeeding. He has some data to indicate that multiple rounds of this process ‘clear out’ autoimmune lymphocytes and help with autoimmune diseases.

Angie Abbacchi
Angie Abbacchi

I’be been Keto for 5 months and after 3 months my triglycerides jumped from 151 to 170, TC from 293 to 354, HDL from 54 to 62. In this 3 month span I added coffee and mct oil daily and until well after noon each day. Could the coffee and the mct oil be a factor in my triglycerides increasing? I have removed the mct oil and will retake labs in a month. Any information would be greatly appreciated.