The Study Keto Influencers Used to Say LDL Doesn’t Matter — Then It Got Retracted

They studied 80 people with dangerously high cholesterol on keto and declared their arteries were fine. They just forgot to mention the soft plaque. Then the follow-up got retracted. Then they tried to save it with a preprint. This is the study I was relying on when my coronary artery was 80% blocked.

After my stent, the first thing I did was go back to the studies I’d relied on.

Not the blog posts. Not the podcast clips. Not the Twitter threads from PhDs who’d never placed a stent or run a code. I went back to the actual papers. The ones I’d skimmed years ago and trusted because someone I respected told me they proved that LDL doesn’t matter if your metabolic markers are clean.

I came to keto in 2017 through the back door — not from influencer culture, but from studying mitochondrial theories of cancer and the Warburg effect. It was a year after a completely clean coronary CT angiogram at age 39, and the metabolic science was genuinely compelling. I was simultaneously becoming interested in the biohacking world due to my background in the business world as a CEO in the health and wellness industry. It seemed fortuitous that the low carb keto / semi carnivore diet I was hearing about on podcasts would end up being something I studied at work to help combat cancer. Needless to say the diet didn’t work for me. So in I dug, looking for that missing signal in the noise.

I needed to know if the science I’d built my diet around had actually said what I thought it said. Because while I was trusting those conclusions, unbeknownst to me, my right coronary artery was quietly closing. My hsCRP was 0.45 – no signs of inflammation. No insulin resistance per labs or the Levels Health glucose monitor. No Lp(a) issue. By every metric the keto and carnivore community uses to wave away an LDL of 160, I should have been fine.

I wasn’t fine. I was one bad day away from a heart attack at 44.

So I started with the study everyone was sharing.

The 2024 KETO Trial: What It Actually Showed

In August 2024, the KETO trial ¹ made waves. Eighty people on ketogenic diets with a mean LDL-C of 272 mg/dL — more than double the recommended level — showed no more coronary plaque than matched controls with an LDL of 123. The keto community celebrated. Case closed. LDL doesn’t matter.

Not so fast.

The 2024 study only reported total plaque scores and calcium. It didn’t break down what kind of plaque was there. That distinction is everything. Calcified plaque is old, stable, and rarely the thing that kills you. Soft plaque — also called non-calcified plaque — is the kind that builds silently in your artery wall, ruptures without warning, and leads to your family standing over your casket wondering what happened – and yes – it happens to people in their 40s and 50s as I see almost every day in the emergency room and as I almost experienced myself. Anyway, in this study they had the imaging technology (CCTA) to differentiate soft (killer) vs hard (calcified) plaque. But the full breakdown of soft vs. hard plaque was not reported in the primary analysis — a notable omission for data they had.

This is the study I was relying on when my LDL hit 160 and I didn’t blink. And also when I felt some gnawing chest tightness at times that I wrote off as anxiety or back and shoulder pain from years of weight lifting. When my doctor mentioned I should consider a statin, I politely declined, as in my mind mild to moderate LDL elevation was “ok” and every podcast I listened told me that metabolically healthy people don’t need to worry about LDL.

The Follow-Up That Flipped the Narrative

In 2025, the same research team published longitudinal data on 100 participants ². This time they used AI-guided quantitative plaque analysis. And the results told a very different story: non-calcified (soft) plaque volume increased by a median of 18.9 mm³ in just one year. In a published reply to criticism of the study ⁷, the authors confirmed this represented a 42.8% relative increase. In a lean, metabolically healthy cohort. Going the wrong direction.

Here’s what makes that number hard to dismiss. This wasn’t a sick population. These were lean, metabolically healthy people with favorable inflammatory markers — exactly the profile the keto community says is protected. And their soft plaque was growing at a rate that would concern any cardiologist, in a cohort that had no proper longitudinal control group receiving serial CCTA scans. Without that control, you can’t even contextualize how alarming the progression is, because the study wasn’t designed to let you compare it. That’s not a detail. That’s a design limitation that should have been front and center.

But there was a deeper problem. The study preregistered soft plaque volume change as its primary outcome. Instead, the published paper emphasized PAV — percent atheroma volume — a secondary endpoint. The raw soft plaque volume change was reported, but the 42.8% relative increase wasn’t. The authors later acknowledged that omission in their reply letter ⁷, calling it “a sincere oversight, not intentional selective reporting.” That’s two papers in a row where the most clinically relevant soft plaque data was either absent, incomplete, or buried behind a different metric.

That soft plaque was building in my arteries at a rate I couldn’t feel, couldn’t test for with standard screening, and didn’t know to look for — because the study that should have raised the alarm was too busy telling a different story.

Then It Got Retracted

Here’s where most people sharing this research on social media haven’t caught up yet.

The 2025 Soto-Mota paper was retracted. Not corrected. Not amended. Retracted — marked as unreliable by the journal itself. The journal cited methodological concerns “too great to be corrected with a corrigendum” — meaning the problems were so deep that even a formal correction couldn’t save it. That’s not a data entry error. That’s the journal saying the paper can’t be trusted.

Think about what this means for anyone who read the original study, saw the headline “Plaque Predicts Plaque, ApoB Does Not,” and used that to justify ignoring their lipid panel. So to restate, the paper that claim came from was retracted, but unfortunately still exists on Pubmed.

The Attempted Rehabilitation

In January 2026, Budoff and colleagues posted a preprint on medRxiv — a reanalysis of the same cohort using different plaque quantification methods (QAngio® + HeartFlow AI-CPA). This is the team returning to the same data with different analytical software.

I want to be precise about what this is. A medRxiv preprint is a manuscript posted publicly before peer review. It hasn’t been vetted by independent reviewers. It hasn’t been accepted by a journal. It’s a draft. And there’s a deeper issue: going back to the same dataset with different analytical tools after your original analysis was retracted is called post hoc reanalysis — rerunning the numbers with a different approach until something looks better. It’s one of the most criticized practices in clinical research. The bar for credibility here is higher, not lower, and that bar is peer review, which this preprint hasn’t cleared.

The “ApoB Doesn’t Predict Plaque” Problem

The 2025 study’s marquee claim — that ApoB doesn’t predict plaque progression — is the one I fear is possible to get someone in trouble. Here’s what actually happened.

Soft plaque grew in nearly everyone in the study. Median soft plaque volume went up 18.9 mm³. PAV went up 0.8%. The plaque increased. That part isn’t in dispute.

What the authors found is that within their cohort, the people with higher ApoB didn’t progress faster than those with lower ApoB. And from that, they concluded ApoB doesn’t drive plaque in this population.

The problem? Every single person in this study had massively elevated ApoB. The median was 178 mg/dL. The interquartile range was 149 to 214. For context, the ESC/EAS guidelines consider optimal ApoB to be below 90 mg/dL for moderate-risk individuals ⁸. So the “low ApoB” group in this study was still running ApoB levels that most lipidologists would flag as high-risk.

This is a textbook restriction of range problem. If you took 100 people who all smoke two packs a day and tracked them for a year, you probably wouldn’t find a dose-response relationship between cigarettes smoked and lung damage either. Not because cigarettes are safe, but because there’s no meaningful variability in your exposure variable. Everyone is already in the danger zone. The study can’t detect what it’s trying to test.

To actually prove that ApoB doesn’t drive plaque, you’d need a control group with normal ApoB levels getting the same CCTA scans over the same period. They don’t have that. What they have is 100 people with uniformly high ApoB all progressing at roughly similar rates — which is exactly what you’d expect if ApoB is causal and they’re all above the threshold that matters.

And this isn’t a fringe position. The causal relationship between ApoB-containing lipoproteins and atherosclerosis is one of the most well-established in cardiovascular medicine, confirmed by Mendelian randomization studies and decades of evidence across study designs ⁴. A 2019 Mendelian randomization study of over 654,000 participants confirmed that coronary heart disease risk is proportional to ApoB particle number, not LDL-C or triglycerides independently ⁵. And as recently as September 2024, the National Lipid Association published an expert clinical consensus confirming ApoB’s superiority over LDL-C for cardiovascular risk assessment, recommending it in routine clinical management ⁶. When LDL-C and ApoB are discordant (not in agreement), risk tracks with ApoB. That’s not one person’s opinion. That’s the consensus of the field.

One underpowered study with a retracted follow-up doesn’t overturn that. It doesn’t even scratch it.

There’s Also a Survivorship Problem

To be in this study, you had to be alive, still on keto, and willing to participate. Anyone who developed cardiovascular disease and stopped the diet — or worse — wasn’t counted. It’s survivorship bias, and it’s baked into the design.

I know what survivorship bias looks like from the other side of it. I ran codes in the ER on people who never got the chance to be in a follow-up study. They’re not in anyone’s dataset. They’re not sharing their n=1 on Twitter. They’re in the cemetery.

Let Me Be Clear About Something

I’m not here to bash ketogenic diets. There’s legitimate evidence for their use in epilepsy, certain metabolic conditions, possibly cancer in some populations, and weight management for some people. Context matters. What I am here to do is point out when a study is being used to make claims it cannot support, especially when those claims might convince someone to ignore a lipid panel that could save their life.

I believed this research. I built my diet around the framework it supported. And while I was trusting the conclusions, plaque was building in my coronary arteries that no standard test could see and no standard screening would catch. I had a master’s in bioinformatics, published research, 15 years inside the health industry, and I deal with codes and heart attacks all day in the ER — and I still got it wrong. That’s not a failure of intelligence. It’s a failure of the information ecosystem.

If your LDL is elevated and on the rise and you’re banking on one underpowered study — whose follow-up was retracted — to tell you everything is fine, you owe it to yourself to dig deeper. Ask your doctor about ApoB testing. Ask about a CCTA with plaque characterization, not just a calcium score. Don’t wait for symptoms. I had just minor nagging chest and mid back tightness. And, I suggest thinking twice before ignoring your doctor when they suggest a statin. Mayo Clinic’s top interventional cardiologist put a stent in me. The first words when he came to visit me in the recovery room, get your LDL as low as you can. Below 55. I took him at his word and have it down to 28 now, 50% better. And the best news, a fresh CTCA last month shows a wide open stent and regressing soft plaque. I am now winning that war. Did I turn vegan? Definitely not. Do I still eat steak? Sure do, and I still love it. But I gave up keto and I gave up “carnivore light” and I feel great and relieved to know my plumbing isn’t becoming clogged, it’s clearing.

As for the studies I have discussed here, the evidence doesn’t support the claim that extreme LDL elevations on keto are benign. Mine sure weren’t. And I am sure others out there have similar stories, theirs just hasn’t unfolded yet, and hopefully that last chapter isn’t in an emergency room.

What would change my mind

I hold this position with conviction, but I hold it loosely. Three pieces of evidence would update it:

• Decades-long prospective cohort data on LMHR-phenotype individuals showing equivalent or lower CV event rates compared to matched non-LMHR controls. Not one-year imaging studies. Not cross-sectional comparisons. Real long-term hard endpoints over 10 to 20 years. The LMHR hypothesis has not yet met this bar.
• A mechanistic explanation for why ApoB particles in LMHR phenotypes don’t drive atherogenesis the way they do in everyone else. Not “their metabolism is different.” A specific molecular pathway showing why an ApoB particle in an LMHR person fails to interact with the endothelial wall, fails to be retained in the intima, or fails to trigger the macrophage cascade. The lipid energy model explains why the LDL is elevated. It does not explain why that elevation wouldn’t do what elevated LDL always does.
• Mendelian randomization data showing that LDL-lowering variants don’t reduce CV risk in LMHR-phenotype carriers. This would be the cleanest test. It hasn’t been done.

If any of those three pieces of evidence emerged, I would update — publicly, on this site. The path to changing my position is open. It just goes through evidence I haven’t seen yet.

A note on convergence

One updated note worth adding: in April 2026, Peter Attia and Tom Dayspring published a substantially deeper version of this same critique on Attia’s site (“There is no safe gamble with high LDL cholesterol”). They reach the same conclusions independently — the 2024 KETO trial doesn’t support the claims being made from it, the LMHR hypothesis lacks the long-term data it needs to overturn the apoB-causality consensus, and the practical risk asymmetry favors lowering LDL rather than gambling on metabolic exception. Their piece goes deeper into the lipoprotein biology than I do here. The fact that two independent analyses with different angles reach the same conclusion is part of how a field’s consensus actually moves. We’re seeing that here in real time.

What’s Next

Here’s the thing that haunts me most about my own case. My 12-lead EKG was clean. My echocardiogram was normal. My Holter monitor was unremarkable. Every standard cardiac test said I was fine. I needed a stent.

If the studies can’t tell you you’re safe, and an LDL panel alone can’t tell you, what test actually works? Next week, I’m going to walk through exactly why standard cardiac screening missed an 80% blockage — and what I’d demand instead if I were you and talking to your primary care doctor about any new or odd nagging pains, chest tightness, or not as much “pep in your step” during workouts. It might just save your life.

---

For more evidence-based analysis of the claims shaping your health decisions, visit calibratedsignal.com. Hard science, delivered honestly. No sponsors. No cheerleading. Just signal.

---

References

1. Budoff MJ, et al. “Carbohydrate Restriction-Induced Elevations in LDL-Cholesterol and Atherosclerosis: The KETO Trial.” JACC Advances. 2024;3(8):101109. PMID: 39372369
2. Soto-Mota A, Norwitz NG, et al. “Longitudinal Data From the KETO-CTA Study: Plaque Predicts Plaque, ApoB Does Not.” JACC Advances. 2025;4(7):101686. PMID: 40192608 [RETRACTED]
3. Budoff MJ, et al. “The Impact of Sustained LDL-C Elevation on Plaque Changes.” medRxiv. Posted January 15, 2026. doi: 10.64898/2026.01.15.26343955 [PREPRINT — NOT PEER-REVIEWED]
4. Ference BA, et al. “Low-density lipoproteins cause atherosclerotic cardiovascular disease. 1. Evidence from genetic, epidemiologic, and clinical studies.” European Heart Journal. 2017;38(32):2459-2472. PMID: 28444290
5. Ference BA, et al. “Association of Triglyceride-Lowering LPL Variants and LDL-C-Lowering LDLR Variants With Risk of Coronary Heart Disease.” JAMA. 2019;321(4):364-373. PMID: 30694319
6. Soffer DE, et al. “Role of apolipoprotein B in the clinical management of cardiovascular risk in adults: An Expert Clinical Consensus from the National Lipid Association.” J Clin Lipidol. 2024;18(5):e647-e663. PMID: 39256087
7. Soto-Mota A, et al. “Reply: Longitudinal Data From the KETO-CTA Study.” JACC Advances. 2025;4(7):101862. PMID: 40450909
8. Mach F, et al. “2019 ESC/EAS Guidelines for the management of dyslipidaemias: lipid modification to reduce cardiovascular risk.” European Heart Journal. 2020;41(1):111-188. PMID: 31504418

---

Nick Hanson, MS, RN, CEN
Mayo Clinic Board Certified Emergency Department RN
MS Bioinformatics & Computational Biology
Duke University APRN-FNP Candidate
Published Epigenetics and Oncology Researcher
Former Health & Wellness Industry CEO (15+ years)
Certified Personal Trainer (ISSA)