Why did my LDL-C rise 22% during the fastest phase of weight loss?

Twelve weeks in, losing roughly 1.1 kg a week, and my fasting LDL-C has gone from 2.7 to 3.3 mmol/L. Total cholesterol up from 4.6 to 5.2. Triglycerides down from 1.9 to 1.4, so the triglyceride story is going the way it should. I am eating considerably less of everything, including saturated fat, so a 22% rise in LDL-C makes no sense to me at all.

I also had an unremarkable ultrasound two weeks ago that mentioned "sludge" in the gallbladder, which I gather is common during rapid loss, and I do not know whether the two findings are connected or coincidental.

Things I have considered and cannot rule out: the draw was after a bad GI week so I may have been dehydrated; I have been eating far more eggs than before because they are one of the few things that goes down easily; and the panel was run at a different lab from the baseline. But none of those feels big enough to explain 22%.

Is there a real physiological reason cholesterol rises while you are actively losing fat, and if so, does it come back down?

The Friedewald issue deserves working through properly, since it is the mechanical part of the answer and you can check it yourself with numbers you already have.

The equation and its failure modes

LDL-C = TC − HDL-C − (TG ÷ 2.2)      [mmol/L]
LDL-C = TC − HDL-C − (TG ÷ 5)        [mg/dL]

The divisor encodes an assumption: that the ratio of triglyceride to cholesterol in VLDL particles is fixed at about 5 to 1 by mass. It is not fixed. It varies with triglyceride concentration, with insulin sensitivity, with fasting state and with the composition of the remnant pool — all four of which you are actively changing.

Three consequences:

• The equation is invalid above a triglyceride of about 4.5 mmol/L (400 mg/dL) and most labs will refuse to report a calculated LDL-C above that. Below it, the error grows steadily as triglycerides rise.
• It systematically underestimates LDL-C at high triglycerides and at low LDL-C. Someone with a calculated LDL-C of 1.6 mmol/L and triglycerides of 3.0 may have a true LDL-C well above that.
• Because it subtracts a triglyceride-proportional term, any fall in triglycerides raises the calculated LDL-C even with an unchanged true value.

Your numbers

Assume an HDL-C of 1.15 mmol/L at both draws for illustration. Then:

• Baseline: TC 4.6 − 1.15 − (1.9 ÷ 2.2) = 4.6 − 1.15 − 0.864 = 2.59. Close to your reported 2.7.
• Follow-up: TC 5.2 − 1.15 − (1.4 ÷ 2.2) = 5.2 − 1.15 − 0.636 = 3.41. Close to your reported 3.3.
• Now recompute the follow-up as if triglycerides had not changed: 5.2 − 1.15 − 0.864 = 3.18.

So of the 0.82 mmol/L apparent rise in LDL-C, about 0.23 mmol/L — roughly 28% of it — is the triglyceride term shrinking rather than LDL cholesterol increasing. The rest is real, and is the mobilisation effect described in the accepted answer.

The measures that do not have this problem

• Non-HDL-C = TC − HDL-C. Baseline 4.6 − 1.15 = 3.45. Follow-up 5.2 − 1.15 = 4.05. A rise of 17%, cleanly measured, with no assumption about VLDL composition. This is your real signal.
• ApoB. A direct immunoassay count of atherogenic particles. No equation, no fasting requirement, no triglyceride dependence.
• The Martin-Hopkins estimate, which replaces the fixed divisor of 2.2 with a value drawn from a 180-cell table indexed by triglyceride and non-HDL-C. Some labs report it by default now. It is materially more accurate than Friedewald at low LDL-C and high triglycerides, and if your lab offers it, prefer it.

General rule worth adopting: track non-HDL-C as your primary calculated number and calculated LDL-C as a secondary one. Non-HDL-C requires no assumptions, is available on every panel already run, and moves for real reasons only.

Adding the "when should I actually draw this" answer, because the timing question is the practical one and it has a defensible answer.

The lipid pool has three distinct states during a weight-loss course and only one of them is worth measuring:

• Baseline, before the first dose. Worth drawing, precisely because it is the only clean pre-intervention state you will ever have. Fasting or not matters little; standardise it and record which.
• Active loss. Not worth drawing for LDL-C or total cholesterol, for all the reasons above. Triglycerides in this window are informative in the sense that they fall early and reliably, but the reference change value for triglycerides is around 60%, so only a large fall means anything from a single pair.
• Stabilised, at least 8 to 12 weeks after weight has plateaued. This is the state that corresponds to the trial-reported changes and to whatever your maintained lipid profile is going to be. Draw here, with ApoB, and treat it as the number that matters.

"Plateaued" needs a definition or it becomes an excuse to never draw. A workable one: weekly weights varying by less than about 0.3% of body weight across four consecutive weeks, with no dose change in that period.

Two secondary notes.

Fasting is now optional for most purposes. Non-fasting panels are acceptable for total cholesterol, HDL-C, non-HDL-C and ApoB, and the triglyceride difference between fasting and non-fasting is typically modest — on the order of 0.2 to 0.3 mmol/L. The exception is if you are being assessed for a triglyceride disorder, where fasting still matters. Fasting for twelve hours while on a drug that suppresses appetite is unpleasant and mostly unnecessary; consistency between draws matters more than the fasting state itself.

And on the gallbladder: sludge on ultrasound in the absence of symptoms is a finding, not a diagnosis, and it frequently resolves. It is nonetheless a reason to know what biliary colic feels like and to have a low threshold for seeking assessment if right upper quadrant pain appears after a fatty meal, because gallstone disease genuinely is more common during rapid weight loss regardless of the mechanism producing the loss. That is a clinician conversation, not a forum one.