I wanted to highlight a newly published paper that’s been gaining a lot of social media attention: Plaque Begets Plaque, ApoB Does Not: Longitudinal Data From the KETO-CTA Trial.

This study centers its focus on so-called Lean Mass Hyper-Responders (LMHRs), which are people who have very high levels of LDL-C but have a normal body weight and are in good metabolic health.

For the past several years, Dave Feldman has been “cautiously optimistic” that the unique metabolic environment of LMHRs might protect them from the cardiovascular risks we’d normally associate with elevated LDL.

This was a hypothesis that I supported. In fact, just a few weeks before this study was published, I joined Dave on his podcast to share my perspective:

Heart disease is complex and multifactorial. Although apoB (the lipoprotein within LDL particles) is causal for the development of atherosclerosis, it is logical to presume that great metabolic health might offset some of the increased risk we’d expect from elevated apoB and LDL.

In other words, perhaps LMHRs are somewhat protected against atherosclerosis despite their elevated apoB and LDL concentrations.

This study, lead by Adrian Soto-Mota and co-authored by Dave Feldman, Nick Norwitz, and Dr. Matthew Budoff is the first to put this idea to the test in a prospective, data-driven way. And for that alone, it deserves major credit.

What Did This Study Set Out to Do?

The Keto-CTA study was originally designed to assess the change in non-calcified coronary plaque volume (NCPV) over the course of one year. This was clearly outlined in the study’s pre-registration (NCT05733325) and reiterated in its published protocol.

Yet, when it came to the actual study publication, this primary outcome was almost entirely neglected.

Instead of focusing on NCPV as a central endpoint, the authors reframed the study’s purpose to emphasize predictors of plaque progression. This shift is plainly visible in the language of the published paper, and you can even see it in the study’s central illustration, where the objective seems to have quietly morphed from “measuring plaque” to “understanding what predicts it.”

We’ll dig into this more when we talk results, but it’s worth pausing here. Downplaying a preregistered primary outcome in favor of post-hoc analyses is a big red flag, especially if the main finding didn’t support the authors’ expectations. It’s the kind of selective reporting that peer review is supposed to catch.

Beyond NCPV, the authors also reported on a few other plaque-related parameters like coronary artery calcium (CAC) scores and the percentage atheroma volume (PAV), as well as associations between blood lipids and these plaque markers.

To be eligible to participate, people had to meet stringent inclusion criteria:

• On a ketogenic diet for over 2 years
• Showing a ≥50% increase in LDL-C after starting the diet
• LDL-C ≥190 mg/dL on keto, but ≤160 mg/dL before
• HDL-C ≥60 mg/dL
• Triglycerides ≤80 mg/dL
• HbA1c <6.0%, fasting glucose <110 mg/dL
• hsCRP <2 mg/L
• Free of hypertension, type 2 diabetes, clinical hypothyroidism
• Not using any lipid-altering drugs or supplements
• Without genetically confirmed familial hypercholesterolemia

This is arguably one of the cleanest metabolic cohorts ever assembled to study this question. These were LMHRs in peak form — lean, with pristine markers of metabolic health, apart from their sky-high LDL-C (which ranged from 202 to 308 mg/dL).

Interestingly, it took years to enroll just 100 participants. That hasn’t been talked about much, but it’s worth noting because many LMHRs are probably not as healthy as they believe themselves to be, which is why recruitment for this study was so difficult.

Dr. Spencer Nadolsky, who originally helped design the study, made this exact point on X.

Unfortunately, there were no control groups in the Keto-CTA study. There were early plans to recruit metabolically healthy adults without elevated LDL-C for comparison (as confirmed by Dr. Nadolsky), but that never happened, possibly due to funding constraints. This absence greatly limits how much we can say about relative risk.

Anyway, let’s look at what the study did (and didn’t) show.

LMHRs Show Accelerated Progression of Unstable Plaque

Let’s start with the primary outcome: non-calcified coronary plaque volume (NCPV).

This was chosen as the primary endpoint for good reason. NCPV refers to the volume of soft, unstable plaque within the coronary arteries that hasn’t yet hardened into calcium. Unlike calcified plaque, which is more stable and easier to detect, non-calcified plaque is more likely to rupture, potentially triggering blood clots, heart attacks, or strokes.

That makes NCPV an early and clinically meaningful marker of atherosclerosis, one that can flag disease before it becomes symptomatic or life-threatening.

Yet despite its importance, NCPV was barely mentioned in the published paper. The only real reference to it appears in Figure 1A, a graphic showing individual-level changes over one year. There are no numbers reported in the text, and the authors only comment that “most participants presented with stable NCPV.”

But when you look at the figure, that interpretation doesn’t hold up. Each line represents a participant, and while a few appear stable, the majority clearly experienced increases in NCPV, especially those who started with higher baseline values.

Clinicians and researchers took note of this and raised concerns on social media. In response, lead author Adrian Soto-Mota shared the actual number that was missing from the paper: the median increase in NCPV was 18.8 mm³—more than 2.5 times higher than the 7 mm³ increase the authors themselves predicted in the study protocol.

So, how does this compare to other populations?

In the PARADIGM study, adults without metabolic syndrome saw an average NCPV increase of 10.3 mm³ over 3.2 years. Those with metabolic syndrome increased by 17.8 mm³ over the same period.

That means the LMHR group had a one-year increase (18.8 mm³) equivalent to what someone with metabolic syndrome would typically develop over three years.

In another study co-authored by Dr. Matthew Budoff (also a co-author on Keto-CTA), adults with dyslipidemia and hypertension (but no diabetes) had an NCPV increase of 28.9 mm³ over 3 years, or ~9.6 mm³ per year. People with type 2 diabetes in that same study saw a change of 52.3 mm³ over 3 years, or ~17.4 mm³ per year.

Again, LMHRs clocked in at 18.8 mm³ per year, which is on-par with type 2 diabetes.

I brought this up to Dr. Budoff, but have not yet heard a response.

Lastly, in the SMARTool Project, which enrolled participants with smoking, obesity, diabetes, hypertension, and dyslipidemia, the annual NCPV increase in non-statin users was just 6.4 mm³. For statin users, it was only 1.0 mm³. Despite being lean and free of metabolic disease, LMHRs had a plaque progression rate nearly three times greater than these high-risk statin-naive participants.

This kind of context was completely absent from the published Keto-CTA paper. And that’s troubling. The data do not support the idea that the LMHR phenotype is protective against atherosclerosis. On the contrary, it appears to be associated with accelerated progression of unstable plaque — worse than in many metabolically unhealthy groups with lower LDL-C levels.

It’s also worth remembering: many would-be LMHR participants were excluded from the study for failing to meet the strict inclusion criteria. If this was the healthiest subset of the group, what would the data look like for the broader population of LMHRs who didn’t make the cut?

This may help explain why the primary outcome was glossed over. The results didn’t align with the narrative some were hoping for.

LMHRs Show Concerning Growth in Total Plaque Burden

Let’s now turn to another key outcome reported in the Keto-CTA study: percentage atheroma volume (PAV).
PAV measures how much plaque (calcified or not) has built up inside the artery walls, expressed as a percentage of the total vessel volume. It’s an important marker in cardiology trials, often used to gauge whether interventions are slowing, halting, or reversing the progression of atherosclerosis.

We can find the median changes in PAV for the LMHRs and several other studies in the supplemental materials. The authors state, “The median change in PAV was 0.8%, and both NCPV and PAV values were comparable with those observed in other cohorts on both visits.”

Let’s unpack that.

The median increase in PAV among LMHRs was 0.8%. Notably, those with a coronary artery calcium (CAC) score of zero at baseline had a lower progression (0.5%), while those with CAC >100 had a much larger jump of 2.4%.

At first glance, these numbers might not seem unusual. The other studies listed in the supplementary table report annual PAV increases in the range of 0.3% to 1.0%. But here’s the problem: those comparison groups were far less healthy than the LMHR cohort.

As highlighted by Gladstone Brookes on X, the participants in those other studies typically had high rates of hypertension, obesity, diabetes, dyslipidemia, and much higher baseline CAC scores. And yet, despite being metabolically compromised, their rates of plaque progression were similar to or lower than the LMHRs, who were supposed to represent a “low-risk” population.

To put this in perspective, in the study by Nakanishi et al, the average PAV increase was 0.7% even though only 13% of participants had a CAC score of zero and 69% had CAC scores above 100. That 0.7% increase is over three times lower than what LMHRs with CAC >100 experienced in the Keto-CTA study (2.4%).

In a subgroup analysis of the PARADIGM study, looking at over 1,000 adults without known coronary artery disease, PAV progressed by 0.45% in those at low 10-year atherosclerotic cardiovascular disease risk, by 0.58% in those at intermediate risk, and by 0.99% in those at high risk.

The healthiest LMHRs, who had CAC scores of zero and fell into the low risk category, still showed a 0.5% increase in PAV, which is on par with or worse than their “less healthy” peers. Meanwhile, LMHRs with CAC scores above 100 saw PAV increases 2.4 times greater than the average high-risk group.

There’s no way to put a positive spin on this.

The Keto-CTA study strongly suggests that LMHRs experience plaque progression at rates equal to or greater than metabolically unhealthy individuals, despite being lean, insulin-sensitive, and inflammation-free. In fact, in those with higher CAC scores, progression was substantially worse than in comparator groups with similar or even more severe cardiovascular risk profiles.

And we have to remember: these were the healthiest LMHRs the researchers could find. Many others were screened out for not meeting strict metabolic health criteria. It’s fair to assume that those excluded would likely show even greater risk if studied.

Taken together, the PAV and NCPV findings strongly suggest that the LMHR phenotype does not protect against atherosclerosis. If anything, it may be associated with accelerated plaque progression, even in the absence of traditional risk factors.

About That “Heterogeneity”

One of the more confusing responses from the Keto-CTA authors on social media has been their strong emphasis on the heterogeneity within the LMHR cohort. In simple terms, they’re pointing out that while some individuals experienced rapid and significant plaque growth, others did not.

As Dave explained on X, they have another follow-up paper in the works to explore why certain LMHRs progressed faster than others.

I think that’s a valuable and necessary line of inquiry. Understanding why some people are more susceptible than others is how we move toward individualized risk assessment and better clinical decision-making.

But let’s not miss the forest for the trees.

This focus on variability feels like a distraction from the core finding of the study: on average, LMHRs experienced accelerated plaque growth. The mean and median values don’t lie. As a group, they are not protected from atherosclerosis.

It’s like arguing that smoking isn’t harmful because some heavy smokers live long, heart-disease-free lives. Yes, outliers exist in every population. But that doesn’t negate the overwhelming evidence that smoking, as a whole, increases cardiovascular risk.

The same logic applies here.

So sure, some LMHRs might have unique protective factors like their genetics, immune profile, lipoprotein subtypes, etc. And I’m genuinely interested in future research that might help identify who’s more or less vulnerable.

But until then, the current data clearly suggest that the LMHR phenotype correlates with rapid progression of atherosclerosis. That should not be ignored or downplayed, no matter how much individual variability exists within the group.

No Link Between Plaque Progression and LDL-C or Apo-B?

Despite all the concerning findings around plaque progression, the authors chose to spotlight something else entirely: the lack of association between changes in plaque volume and apoB levels, which is a protein widely accepted as causal in the development of atherosclerosis.

This idea is what inspired the paper’s title, Plaque Begets Plaque, since baseline plaque burden (like CAC scores) was a stronger predictor of progression than lipid levels.

And to be fair, that is an interesting observation.

But here’s the catch: every participant in the study had very high LDL-C (202–308 mg/dL) and apoB (149–214 mg/dL) concentrations. So the authors are looking for associations within a narrow band of already-elevated values. That’s like trying to study the effects of smoking by only enrolling people who smoke 2 to 4 packs a day.

If you found no correlation between smoking intensity and disease in that group, would you conclude that cigarettes don’t cause harm? Of course not. You’d say, “We need to compare these smokers to non-smokers.” That’s the same problem here.

The Keto-CTA study lacked a control group of metabolically healthy adults with normal or low apoB and LDL levels, which means it can’t address the broader question of whether high apoB drives atherosclerosis. Without that comparison, there’s no way to determine whether high apoB was the reason plaque progressed so rapidly in these LMHRs.

In fact, it’s highly plausible that apoB did play a causal role in the accelerated plaque growth seen in this cohort. These participants were deliberately selected for their clean metabolic health, so what else could explain their elevated risk?

This is a question that still needs investigation, but if anything, the Keto-CTA data reinforces the concern that extremely high apoB levels may override other signs of metabolic health when it comes to cardiovascular risk.

Final Thoughts

I have to give a big applause to Dave and his colleagues for making this study happen. It’s no small feat to follow 100 people for a year and provide this kind of insight into their cardiovascular disease risk, especially for a population that has been widely discussed but never studied.

That said, as I stated on X, the take-home message from the Keto-CTA study is hard to ignore: the LMHR phenotype is not protective against atherosclerosis. On the contrary, it appears to be associated with greater plaque progression than we’d expect, even when metabolic health is otherwise pristine.

Yes, some individuals in the study did show minimal plaque growth, and future research will hopefully reveal what distinguishes those who are protected from those who aren’t. But as it stands now, the LMHR phenotype characterized by very high LDL-C and apoB in the context of leanness and metabolic health should be viewed as a cardiovascular risk factor.

Just like smoking and diabetes, there’s heterogeneity in outcomes. Not everyone who smokes gets heart disease. Not everyone with type 2 diabetes develops atherosclerosis. But that doesn’t mean we stop viewing those conditions as risks.

The same logic applies here.