Starting tomorrow, August 30th, I’ll be executing a replication of the OxLDL experiment I performed last year.

Experiment Design

As with the first, I’ll have three phases:

1. Eat baseline diet for five days (maintenance calories)
2. Eat 1/2 baseline diet for five days (hypocaloric)
3. Eat high calorie keto for five days (hypercaloric)

Additional Notes:

• In the original experiment I sought to have 2x baseline diet but found it untenable and thus made an adjustment. This experiment will attempt to replicate that adjustment as well to keep all food the same.
• My current sleep schedule has me waking up a bit earlier in the morning than last year. Thus I’ll be adjusting both the eating and blood draw windows. (See below)
• As with the original, I’ll be eating at three time periods a day, each five hours apart. One difference is that my scheduled times will be about 1 hour earlier at 9am, 2pm, and 7pm.
• Given the earlier eating times, I’ll be likewise getting my blood drawn at approximately one hour earlier as well.

Blood tests for mornings of September 4th, 9th, and 14th

One exciting change is that I’ll be getting much, much more expansive blood tests than last time:

Labcorp:

• Apolipoprotein A-1
• Apolipoprotein B
• C-Reactive Protein
• Complete Blood Count (CBC)
• Comprehensive Metabolic Panel (CMP)
• Cortisol
• Fatty Acids, Free (NEFA)
• Ferritin, Serum
• Fructosamine
• GGT
• Glucagon, Plasma
• GlycA
• Hemoglobin A1c
• IGF-1
• Insulin and C-Peptide
• Leptin
• Lipid Panel
• Lipoprotein(a)
• Lp-PLA2 Activity
• Nuclear Magnetic Resonance (NMR)
• Oxidized Low-density Lipoprotein (OxLDL)
• Testosterone, Serum
• Thyroid Panel
• Uric Acid, Serum
• Vitamin B12 and Folate
• Vitamin D, 25-Hydroxy

Boston Heart Diagnostics:

• Adiponectin
• Cholesterol Balance
• Fatty Acid Balance
• HDL Map
• Interluekin-6
• Leptin (Redundant, but of high interest)
• Oxidized Phospholipids on apoB (OxPL)

Hypothesis

As with the first experiment, I predict my OxLDL levels will track with LDL particle count (LDL-P).

However — of much greater importance are the results of the Oxidized Phospholipid test (OxPL). If the assay is truly quantitative to this metric, it stands to reason OxPL will not quantitatively track with OxLDL, even if there were a correlation at a smaller magnitude. On the other hand, if it did track in magnitude, this would open up a lot more questions (that I won’t cover here), particularly if OxPL rises substantially in the hypocaloric phase and drops substantially in the hypercaloric phase.

Although involving a lot of unpleasantness, the last three days have been interesting, to say the least. To explain why, we should start from the beginning…

Inspiration for the EVOOvsButter Experiment

Before this experiment, there have been a number of others that used a ketogenic meal replacement mix with different sources of fat. These were mainly carried out by Keto Chow owner, Chris Bair — often in collaboration with yours truly.

The most impactful one for me was when Chris performed a full six weeks swapping in and out various oils that were either predominantly saturated, monounsaturated, or polyunsaturated. (You can read up on it in full here). He had likewise followed up with some further experiments involving friends and family to provide additional data as well.

When doing our “Tandem Drop” experiment, Siobhan used exclusively Keto Chow shakes with heavy whipping cream throughout without issue. In fact, she was getting up to around 6,000 calories a day for the final stretch (note her maintenance calories are closer to 1,500!)

I mention all this because my considering the meal replacement shake with a refined fat source actually has a lot of precedent and appears safe enough, at least in the short term.

Extra Virgin Olive Oil vs Butter – First Attempt

I laid out the design for the original experiment here and got started Tuesday, June 30th, with the Extra Virgin Olive Oil (EVOO) phase. The first three and a half days went okay overall without any substantial symptoms, save some lethargy and and lower appetite (the last shake of each evening took me a while to finally finish off).

However, on the evening of the fourth day I seemed to be feeling a hightened sense of anxiety that I’d describe as comparable to what you get with nausea — but without the nausea. It was mild, but did increase my anxiety and made it difficult to sleep.

On the morning of the fifth day (4th of July), I had a collection of mild symptoms that included less energy, an odd sensation in my feet and calves, and further anxiety that seem to carry over from the night before. Given the leg problems, I was inclined to assume these may be electrolyte issues — something I’ve struggled with before while keto.

I decided to pause the experiment to retool with a stronger electrolyte base for the next version.

Extra Virgin Olive Oil vs Butter – Second Attempt

The revised version of this experiment began three days ago with a much stronger electrolyte regime. I also added in a direct source of protein through consumption of a lean burger patty with every shake. Another key change was a lower quantity of the fat source (i.e. EVOO at 12 tbsp/day vs the original 15)

While the initial Thursday went by without incident, I was feeling symptoms in the morning of the Friday after with both low energy and very mild nausea that appeared to dissipate after a couple hours.

My first scheduled exercise of the day is a 2.3 mile walk at around 7am, where I was especially rundown. Following this I typically do some maintenance strength training ad libitum (such as pushups) which I had no energy for and simply skipped.

For the first scheduled meal at 9am I found I had no appetite, something very unusual. I managed to slowly get through my burger patty but kept eyeing my EVOO shake suspiciously. I was getting more doubtful I could consume it without forcing it down and between this and the collection of symptoms, I reluctantly decided to pause the experiment — for a second time.

While I was optimistic this momentary second pause would snap me back into place, I was wrong. The nausea and lack of energy both increased marginally into the evening and I fell asleep for two hours from 6pm to 8pm on the couch. I then had trouble getting to sleep and kept moving in and out of a state of insomnia (also very unusual).

After getting up on Day 3 (yesterday), I realized I wasn’t anywhere close to normal. While the nausea was gone, my energy levels were even lower and I had no appetite until the afternoon. I felt like one does after having been sick for a while and was now coming out of it — being “on the mend” as it were.

Why So Different This Time?

Much of the reason I’m giving the full accounting of everything before this point is to emphasize the unusual nature of this last attempt. Sure, a clear common denominator is the EVOO and perhaps even the 1/3 dose on my coming revision is still too much, I don’t know yet. But I can’t help but wonder if there’s something different about this time compared to the last.

One thought that occurred to me is that it might have been a bad batch of EVOO. While all of them came from the same Kirkland bottles, the current shakes I made were from a newly opened one in the batch. Is it possible they could be different from one bottle to the next in a material way when consumed at scale?

The Next and Last Attempt

As mentioned above, I’m going to attempt to do this experiment one more time, but with 1/3rd the quantity for each intervention (just 4 tbsp / 57g). I’ll be filling the remaining calories with “neutral foods” that try to hit a 50/50 balance between SFA vs MUFA/PUFA.

I had hoped to be on track with that experiment right away, but clearly I should take a little more time to get re-centered first.

The Toll of Setbacks

One aspect to these ambitious n=1 experiments I don’t talk about is the mental preparation. Kind of like gearing up for a marathon-like competition, I’m spending weeks to months getting my headspace aligned for a long haul of a very structured days. It’s a lot of anticipation and planning depending on the scope of the project — and my experiments rarely get larger than this.

Thus, these kinds of developments that actually disrupt a 4-week planned initiative are very emotionally taxing, particularly given all the meetings, podcasts, etc that I schedule around them. Often my wife will plan meals and experiences for immediately following the completion of these experiments as she knows it can’t happen during, so this news is very disappointing to her as well. (Have I mentioned my wife is a saint for putting up with all this?)

Data is Data

While I have outlined a lot of negatives above, let’s not lose sight of the silver lining — even disappointing results can be valuable data in their own regard. I have some new hypotheses regarding what may have happened that may potentially show up in the coming bloodwork once I have this experiment back underway. But I’ll save those thoughts for a later time.

[Update 2020-08-14: After some issues over the second day (see social media discussion), I’m changing the design and restarting the experiment with lower overall dosage of both the EVOO and butter arms. The schedule is likewise pushed back exactly one week. All changes reflected below…]

The original version of this experiment was unfortunately paused due to some issues that I believe were electrolyte related. I’ve been looking to restart it under a new design but had a number of scheduling conflicts and some unusually low triglyceride readings for the last few weeks that have since normalized.

Experiment Design

This experiment will be a double crossover for five weeks. I will start with a neutral baseline diet, then begin alternating between each intervention for one-week phases. I will attempt to keep all other variables as equivalent as possible throughout: eating times, exercise times/duration, and sleep schedule.

For the extra virgin olive oil (EVOO) I’ll be using Kirkland, and for the butter, Kerrygold. Both will be combined with warm water by emulsion blender and then mixed with a meal replacement shake powder (Ketochow).

Schedule Outline

1. August 13-19 – Baseline
2. August 20-26 – EVOO
3. August 27-September 2 – Butter
4. September 3-9 – EVOO
5. September 10-16 – Butter

Everyday Constants

All meals will consist of one lean beef burger patty from Metabolic Meals.

For the morning meal (~9am) I’ll be adding one or more foods with the intent of meeting close to a 50/50 ratio of saturated fat vs mono/polyunsaturated.

For the middle and late meals (~2pm and ~7pm), I will be consuming roughly 250 calories each (500 total) of either intervention for that week.

Omega 3 such as EPA and DHA will be provided via daily supplements from Zhou.

To prevent against the electrolyte issues from last time, I’ll be consuming 9 grams of pink salt dissolved in water each morning, along with electrolyte beverages ad libitum throughout the day.

Exercise will consist of two to three miles a day of walking with moderate, ad libitum upper body exercise (such as pushups).

Sleep will be ad libitum some time between 10pm and 8am.

Blood tests for mornings of August 20, 27, and September 2, 9

• Apolipoprotein A-1
• Apolipoprotein B
• C-Reactive Protein
• Complete Blood Count (CBC)
• Comprehensive Metabolic Panel (CMP)
• Cortisol
• Fatty Acids, Free (NEFA)
• Ferritin, Serum
• Fructosamine
• GGT
• Glucagon, Plasma
• GlycA
• Hemoglobin A1c
• IGF-1
• Insulin and C-Peptide
• Leptin
• Lipid Panel
• Lipoprotein(a)
• Lp-PLA2 Activity
• Nuclear Magnetic Resonance (NMR)
• Oxidized Low-density Lipoprotein (OxLDL)
• Reverse T3
• Testosterone, Serum
• Thyroid Panel
• Uric Acid, Serum
• Vitamin B12 and Folate
• Vitamin D, 25-Hydroxy

I’m also excited to announce I’ll be adding new tests from Boston Heart Diagnostics as well. These include:

• Adiponectin
• Cholesterol Balance
• Fatty Acid Balance
• HDL Map
• Interluekin-6
• Leptin (Redundant, but of high interest)
• Oxidized Phospholipids on apoB (OxPL)

Endpoints of Interest

As with my original design of this experiment I have outlined two categories of interest: lipid levels and inflammation markers. However, I’ll now be adding an additional test via Boston Heart that I’ve been waiting for — the Oxidized Phospholipids on apoB test (OxPL).

Something I’ve long speculated on is whether OxLDL would track tightly with OxPL. OxLDL is a pass/fail test — either the LDL particle has detectable levels of oxidation or it doesn’t. But the OxPL should show the degree of oxidation in those particles detected as an average for the total sampled. Thus, I think it will be an excellent test for risk and give us much more valuable information (but I do have some caveats, discussed below).

• Will EVOO interventions have lower relative total and LDL cholesterol levels (TC & LDL-C)?
• Will EVOO interventions have a greater oxidized LDL to total LDL particle count ratio (OxLDL/LDL-P)?
• Will EVOO interventions have a greater oxidized phospholipid to oxidized LDL particle ratio (OxPL/OxLDL)

Discussion

As mentioned above, there are many factors that can influence cholesterol levels, particularly LDL. There is one effect that is rarely discussed in the literature but is of particular interest to me — how much we see particular types of dietary fat result in higher or lower oxidation per LDL particle.

In other words, are we seeing lower levels of LDL cholesterol because LDL particles are getting oxidized and cleared by scavenger receptors at a higher rate? There are many limitations to the experiment in how well it can provide evidence to this answer, but it might open the door.

We can’t easily know the true rate of clearance for OxLDL in vivo, or how much this is impacted by the degree of oxidation per particle. However, oxidized LDL particles are commonly understood to be cleared at a higher rate than unmodified LDL. So it will be meaningful data if we find a higher OxPL to OxLDL ratio in one intervention over the other.

Regardless, there will be quite a bit more data from all the other blood markers to provide comparisons that go well beyond lipids.