May 11

The Oxidized LDL Experiment

For quite a while I’ve been wanting to test a theory I’ve had regarding the blood test for oxidized LDL (oxLDL). The test has some research behind it suggesting a correlation with heart disease, which makes sense given this association of oxLDL in the artery walls and atherogenesis.

There’s an assumption that oxLDL will track with oxidative stress in the body. But I’ve long had the hypothesis oxLDL will track with LDL particle count (LDL-P). For example:

Deeper Mechanistic Reasons

In a previous draft of this blog post I go into my lengthy reasons for why this makes sense to me, but I think I’ll save that for later as some of it is a bit controversial.

But I’ll tease part of the answer can be suggested by this tweet:

Designing the Experiment

So is there a way to increase LDL-P while keeping oxidative stress roughly the same or even lower? Why yes, we use the Inversion Pattern.

  1. Eat baseline diet for five days, wide spectrum test + oxLDL
  2. Eat 1/2 baseline diet for five days, wide spectrum test + oxLDL
  3. Eat 2X baseline diet for five days, wide spectrum test + oxLDL

The hypothesis: oxLDL will track with LDL-P. It will go up with LDL-P goes up and drop when LDL-P drops. If oxLDL has little to no movement or tracks inversely, this would be clear evidence against it.

In other words, if LDL-P increases from phase 1 to 2, and drops from phase 2 to 3, we should see similar movements with oxLDL.

Unexpected Amendment at Phase 3

While I did caveat in advance I might not be able to hit the Phase 3 goal of doing twice my baseline diet, I quickly found out my fear was met. Pretty quickly I realized I couldn’t consume that much food that quickly.

See my video on this as it happened here:

Thus, I compensated the difference with some Keto Chow shakes which allowed me to move up my levels a lot faster. (Full disclosure, Keto Chow provides us product support for experiments, but we draw no financial compensation from the company or have an agreement with it of any kind)

There were also additional circumstantial stressors in that last phase unrelated to the experiment, so I made note of this too and how it may be a confounder as well.

Process Woes

As an aside, getting this testing was both expensive and extra cumbersome. The lab I was getting by regular tests through didn’t allow for Boston Heart blood to be taken, so I had to make other arrangements to get the blood drawn and spun separately on that test. Then I had to pack it in ice myself and take it to FedEx to ship immediately afterward.

As experiments go, this one has a lot of planning and footwork to pull off by comparison.

The Thrill of Anticipation

The first test rolled in and these results would now serve as baseline.

Okay, 88 it is.

While waiting for the second test, I happened to be in the area of the clinic I was getting these labs through and dropped by. “Hey, any chance you got that second oxLDL test?”

“Actually, we got the email notification this morning. We’ll print it out.” Said the administrative assistant.

“Great!” I felt a wave of excitement hit me. “Before I see it,” I began, staring at the printer, “I predict the number will be higher than the test on the 12th before it.”

The sheet finally popped out and the assistant handed it to me. “YES!” I shouted, and then suddenly dialed it back, self-consciously remembering I’m in a doctor’s office. “Sorry.”

Certainly 128 was an incredible change for a five day difference. The effect size was more than I could have hoped for to test the hypothesis.

I was already out of town at the moment the last test came in. When I let the admin know I wanted her to send it over via email, she said, “Awww, we were hoping you’d see it in person here for the reaction.” We both chuckled.

Wow! 75 at the low. And again, this is just five days from the prior phase endpoint at 128.

Now again, I have to be intellectually honest. That last phase could have been confounded by the unexpected stressors and my having to add on Keto Chow to reach the caloric surplus. It could be posited one or both of these further impacted the outcome. Probably a stretch, but worth the mention.


Unsurprisingly, this is the tightest regression line I’ve ever posted.

To be sure, there were many runner ups that had less correlation, yet still in the 90s. Here’s the big rundown of both the diet and blood markers:

Final Thoughts

Needless to say, I was quite happy to finally have this experiment completed and ecstatic at the outcome data. While the correlation with LDL-P was something I’ve predicted for some time, I’ll concede it was far tighter than I was imagining.

I have many other thoughts on oxLDL which I’ll save for another time. But the biggest takeaway is how clearly dynamic this marker is, along with just about every other lipid that is central to the research of this site.

Leave a Reply

Photo and Image Files
Audio and Video Files
Other File Types
6 Comment threads
0 Thread replies
Most reacted comment
Hottest comment thread
6 Comment authors
Nick HallLouisPatricia L ShellyjgilberAZSandr Carusetta Recent comment authors

newest oldest most voted
Notify of

Great experiment!

Sandr Carusetta
Sandr Carusetta

Wow- thanks for going through all of this – and for sharing it! Your educational generosity arms us with practical, decision-making ammo.



oxLDL / LDP-P * 100

12-Apr 4.17%
17-Apr 4.94%
22-Apr 3.85%

Is the ratio more important than the count?

ie, if the ratio was more like 20% oxLDL, that would be a real risk factor, almost regardless of the actual LDL-P, itself?

Patricia L Shelly
Patricia L Shelly

Interesting but not yet sure how to use this info


Hello, I just got my lipid profile results

Total Cholesterol is 481
HDL 67
LDL 387.8
VLDL 26.2
Triglycerides 131

With such high LDL, Is this to be of concern?

Note: I am on a zero carb diet, although I cheat a little here an there. My blood pressure is 92 sys, 58 dia, and my pulse is 64.

Nick Hall
Nick Hall

Thanks for this experiment Dave – fascinating!

I have taken the liberty of plugging your numbers into your remnant cholesterol report form. See the attached graphic.

Obviously the various markers show the same degree of variability as do your basic lipids.

Has this experiment given you any thoughts about the validity of remnant cholesterol numbers versus actual inflammatory risk?

For example, it seems from the numbers that doing caloric reduction after your baseline diet significantly changes (in an undesirable direction) your number for “Atherogenic Index of Plasma”…. (actually the other widely used ratios aee also changed similarly.)

It seems to me that a *really* useful additional biometric in this situation would be, for example, high-sensitivity CRP…. (i.e. one or more biometrics that track actual inflammatory response)