Lean Mass Hyper-Responders See Rapid Plaque Progression

A newly published study – “Plaque Begets Plaque, ApoB Does Not: Longitudinal Data From the KETO-CTA Trial” – has stirred substantial discussion in the nutrition and cardiology communities [https://www.jacc.org/doi/10.1016/j.jacadv.2025.101686]. 

This study examined so-called Lean Mass Hyper-Responders (LMHRs), individuals who develop very high LDL-cholesterol levels on a ketogenic diet despite being lean and in excellent metabolic health. For years, researchers of LMHRs (notably Dave Feldman, who helped coin the “LMHR” term, and Nick Norwitz, PhD) have hypothesized that these unique metabolic conditions might protect LMHRs from the usual atherosclerotic risk of high LDL-C. 

In other words, perhaps being lean, insulin-sensitive, with high HDL and low inflammation could offset the cardiovascular dangers typically associated with elevated apoB/LDL levels. Indeed, an earlier cross-sectional report found that long-term LMHR dieters had coronary plaque levels comparable to matched controls with normal LDL, offering some initial reassurance that diet-induced high LDL might not translate into higher plaque burden, at least at baseline. 

The new KETO-CTA study is the first prospective trial to put this optimism to the test by tracking actual plaque progression in LMHRs over time. The results provide important – and sobering – data on this question.

Table of Contents

• Study Design and Participant Criteria
• Key Findings: Plaque Progression in LMHRs
• How Does This Compare to Other Cohorts?
• Criticisms and Controversies
• Authors’ Rebuttals and Responses
• Final Thoughts

Study Design and Participant Criteria

The KETO-CTA trial followed 100 LMHR (or near-LMHR) individuals on long-term ketogenic diets, tracking their coronary arteries over one year via CT angiography. All participants met stringent inclusion criteria to ensure they were truly metabolically healthy apart from LDL. Key criteria included: at least 2 years on a ketogenic diet; LDL-C ≥190 mg/dL on keto (but <160 mg/dL pre-diet), with a ≥50% LDL increase upon carb restriction; HDL-C ≥60 mg/dL; TG ≤80 mg/dL; HbA1c <6.0%; fasting glucose <110 mg/dL; hs-CRP <2 mg/L; no hypertension, diabetes, or other major comorbidities; and no use of lipid-lowering drugs or supplements.

Anyone with genetically confirmed familial hypercholesterolemia was excluded. In short, this was arguably one of the cleanest metabolic cohorts ever assembled to study diet-induced high cholesterol. The average age was mid-50s, and baseline LDL-C averaged ~237 mg/dL (median, Q1–Q3 202–308) with ApoB ~178 mg/dL – sky-high LDL/ApoB levels paired with “pristine” metabolic health markers.

Notably, it took several years to recruit 100 participants meeting these strict criteria. 

This difficulty was pointed out by clinicians like Dr. Spencer Nadolsky, who noted that many self-identified LMHRs were not as healthy as they thought and thus failed to qualify.

In other words, the study likely represents a best-case subset of LMHRs. Originally, the investigators planned to include a control group of metabolically healthy people with normal LDL for comparison, but this fell through (reportedly due to funding constraints). 

As a result, the study is a single-cohort longitudinal design – 100 LMHRs scanned at baseline and 1-year follow-up – without an internal control arm. This lack of a control group, as we’ll discuss, limits the ability to directly quantify how their plaque progression compares to what would happen without high LDL.

The primary endpoint pre-specified for the trial was the change in non-calcified coronary plaque volume (NCPV) over one year [https://clinicaltrials.gov/study/NCT05733325]. NCPV represents the volume of “soft” plaque in the coronary arteries that has not yet calcified. 

The focus on non-calcified plaque is important because soft plaque is more vulnerable – it’s the type more prone to rupture and cause heart attacks, whereas calcified plaque is more stable. Thus, NCPV is considered an early, clinically meaningful marker of atherosclerosis progression. 

The study also tracked percent atheroma volume (PAV) – a measure of total plaque burden relative to artery volume – and coronary artery calcium (CAC) scores, among other parameters. Blood tests (LDL, ApoB, etc.) were collected to explore correlations between lipid levels and plaque changes. Participants were instructed to remain on their diets, and adherence was monitored. After 1 year, the changes in plaque and the predictors of those changes were analyzed.

Key Findings: Plaque Progression in LMHRs

1. Non-calcified Plaque Volume (NCPV): 

This was the pre-registered primary outcome [https://clinicaltrials.gov/study/NCT05733325], yet surprisingly the published paper provided almost no numerical data on it. The authors only commented that “most participants presented with stable NCPV,” and the main text omitted the actual change in NCPV. 

However, the study’s Figure 1A (showing individual patient trajectories) and subsequent disclosures tell a different story. 

When pressed by clinicians on social media, lead author Dr. Adrian Soto-Mota revealed the missing number: the median NCPV increased by 18.8 mm³ over one year. 

For context, the study’s own protocol had anticipated only about a 7 mm³ increase in NCPV [https://pmc.ncbi.nlm.nih.gov/articles/PMC11426984/]. 

The median progression was ~2.5-fold higher than expected, indicating substantially more soft plaque buildup than the investigators had estimated. 

Indeed, looking at the individual data, the majority of participants experienced an increase in soft plaque, especially those who had any appreciable plaque to begin with. Only a minority showed truly “stable” NCPV, and a few even had small decreases (which could be due to measurement variability). 

Overall, however, the group trend was clearly toward increased soft-plaque volume – a worrisome finding, given that non-calcified plaque is the type most associated with acute cardiac events.

2. Percent Atheroma Volume (PAV): 

This metric captures total plaque burden (both calcified and non-calcified) as a percentage of artery volume. The median PAV increased by 0.8% over one year in the LMHR cohort. At first glance +0.8% PAV might not sound remarkable. But remember, these were individuals with very low baseline plaque (median CAC=0, and median baseline PAV only around ~1.6%). An 0.8% absolute increase represents roughly a 50% relative growth in total plaque burden in one year. 

Moreover, when the authors broke down results by baseline calcium scores, there was a clear gradient: those who started with zero CAC had a median +0.5% PAV (still an increase) while those who had CAC >100 had a whopping +2.4% PAV progression. 

In plain terms, LMHRs who already had some calcified plaque at baseline accrued a lot more new plaque in a single year. Even the “best-case” LMHRs (CAC=0 at start) didn’t avoid plaque growth – they still averaged +0.5% PAV, which is on par with, or worse than, what we’d expect in metabolically less healthy people (as we’ll see below). 

Coronary calcium scores themselves tended to rise in many individuals, though specifics on CAC progression were not highlighted in the paper. Qualitatively, the data suggest that if plaque was present at baseline, it tended to expand over the year, whereas plaque-free arteries mostly remained clear (but even some of those gained small new plaques, contributing to the PAV increase).

It’s important to emphasize the clinical significance: the plaque that did grow was disproportionately non-calcified (soft) plaque. Calcified plaque volume (if measured) was not the focus; instead NCPV expansion drove the PAV changes. Soft plaque is more dangerous per volume than calcified plaque, so an increase in NCPV is not easily dismissed as benign. 

Taken together, these outcomes show that on average, LMHRs experienced significant progression of coronary atherosclerosis over just one year, despite their impeccable metabolic health profiles.

How Does This Compare to Other Cohorts?

One way to gauge the magnitude of these findings is to compare against plaque progression rates reported in other studies (especially since this trial lacked its own control group). The numbers paint a stark picture: LMHRs had plaque growth equal to or exceeding that of much “higher-risk” populations.

• PARADIGM Study (general population, with/without metabolic syndrome): In a large observational study (PARADIGM) of 1,200 adults undergoing serial CTAs, individuals without metabolic syndrome showed an average NCPV increase of ~10.3 mm³ over 3.2 years, whereas those with metabolic syndrome had about 17.8 mm³ over 3.2 years. 

Annualized, that’s roughly 3.2 mm³/year (no MetS) vs 5.6 mm³/year (MetS). The LMHR median of ~18.8 mm³ in one year is comparable to three years’ worth of plaque accrual in people with metabolic syndrome [14]. In fact, an LMHR’s plaque progression in 1 year outpaced the 3-year progression of an average metabolically unhealthy person, and was about triple the rate seen in metabolically normal individuals in PARADIGM.

• Diabetes and Hypertension (Budoff et al. cohort): In another study co-authored by Dr. Matthew Budoff (who is also a KETO-CTA author), patients with dyslipidemia and hypertension but no diabetes had an NCPV increase of ~28.9 mm³ over 3 years (~9.6 mm³/year) [https://pubmed.ncbi.nlm.nih.gov/21163451/]. 

Those with type 2 diabetes had ~52.3 mm³ over 3 years (~17.4 mm³/year). Strikingly, the LMHR group’s 18.8 mm³/year is on par with the progression rate seen in frank type 2 diabetics over the same time frame. In other words, these ostensibly “low-risk” keto dieters accumulated soft plaque as fast as patients with one of the strongest cardiovascular risk factors (diabetes).

• SMARTool high-risk cohort: The SMARTool project studied people with multiple risk factors (obesity, smoking, hypertension, dyslipidemia, diabetes) undergoing serial CTA. Even in this high-risk population, those not on statins had an average NCPV increase of ~6.4 mm³ per year, and those on statins had only ~1.0 mm³ per year [https://www.jacc.org/doi/10.1016/j.jacadv.2024.101459]. 

LMHRs – who were lean, normotensive, non-diabetic, etc. – still had nearly 3× the plaque growth of the high-risk non-statin group [14]. Compared to the statin-treated group, LMHR progression was ~18× higher (since many LMHRs had shunned lipid-lowering therapy). This context underscores how unexpectedly high the LMHR plaque progression was, given their favorable baseline health.

Recall the LMHR median was 0.8%, with subsets ranging up to 2.4%. The PARADIGM registry reported PAV progression stratified by risk: low 10-year risk patients had ~+0.45% PAV/year, high-risk patients ~+0.99%/year. 

Even the healthiest LMHRs (CAC=0) gained +0.5% – which is as much or more plaque than low-risk populations  – and LMHRs with CAC>100 saw PAV increases 2.4 times greater than the average high-risk group [11]. In fact, a CAC>100 LMHR’s PAV increase (2.4%) was ~2.4× the average high-risk person’s. 

This suggests that once a LMHR has developed some atherosclerosis, it may advance very rapidly under continued high-LDL exposure.

As the comparisons illustrate, the LMHR phenotype did not confer protection against plaque buildup – if anything, these individuals accumulated plaque faster than many groups traditionally considered at higher risk. 

It’s an arresting finding: being lean, fit, and inflammation-free did not stop the march of atherosclerosis when LDL was extremely elevated. Recall that the trial participants were the cream of the crop; many less healthy LMHRs never qualified for the study. 

One has to wonder what the plaque progression might look like in the broader LMHR population (e.g. those with borderline blood pressure, or slightly higher inflammation, etc.) – likely even more pronounced. 

All told, the data strongly suggest that LMHRs are not immune to atherosclerosis. On the contrary, their high LDL appears to be associated with accelerated plaque progression, even in the absence of other risk factors.

Criticisms and Controversies

Given the unexpected (and for some, unwelcome) findings, the KETO-CTA paper has been closely scrutinized. Several criticisms emerged on social media and in blog analyses almost immediately:

Selective Reporting of Primary Outcomes: 

Perhaps the most prominent critique is that the authors downplayed their primary endpoint (NCPV) because the result was unfavorable to the “LMHR might be safe” narrative. 

As noted, the paper’s text barely mentions NCPV at all – no mean or median change is reported, and only a passing comment is made that most individuals were “stable”. 

This is puzzling for a pre-registered primary outcome and raised red flags for observers. 

Instead, the published article reframed the focus to predictors of plaque progression (e.g. baseline plaque vs. apoB levels), effectively shifting the spotlight away from the magnitude of plaque increase itself. 

Such a post-hoc reframing can be a form of outcome switching. As nutrition researcher Alex Leaf remarked, “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” [https://alexleaf.com/lean-mass-hyper-responders-are-not-protected-from-atherosclerosis/]. 

The concern is that the investigators may have been “hoping for” minimal plaque growth, and when the data showed the opposite, they instead highlighted a different story (“plaque begets plaque”) that was not the original primary question. 

This kind of selective emphasis is something peer reviewers usually try to catch. In this case, it slipped through, leading to criticism that the published report might be presenting a somewhat rosier interpretation than the raw results justify.

Use of Median vs. Mean (Data Presentation): 

Related to the above, commenters noted that the study reported median changes in plaque rather than means. For skewed data, medians can be appropriate, but the worry here is that by only giving medians (and then not even giving them in the main text), the authors may have obscured the full extent of progression in some individuals. 

For instance, the median NCPV +18.8 mm³ tells us half the group exceeded that and half were below. The wide interquartile range indicates some participants had much larger increases. 

If those were averaged in, the mean change might be even higher, underscoring the risk. 

Moreover, in the trial’s registered protocol, the expected change of 7 mm³ was described as an average (mean) [https://pmc.ncbi.nlm.nih.gov/articles/PMC11426984/].

Comparing a mean expectation to an actual median outcome is a bit asymmetric; had they reported the mean NCPV increase, it might have exceeded 20 mm³, making the contrast with expectations even starker. 

Lack of Control Group: 

As mentioned, without a parallel control arm, it’s hard to say how much of the observed plaque progression was due to the high LDL per se. 

The study demonstrated that LMHRs did develop plaque at a notable rate, but we must infer comparison to normal-LDL people from historical data.

If, for example, a matched group of ketogenic dieters without LDL elevation had been followed, would they have shown near-zero plaque growth? We can suspect so, based on prior studies (and the baseline comparison that showed similar plaque at one point in time). 

The lack of a contemporaneous control means the study cannot conclusively answer the causal question – i.e. would these people have avoided plaque progression if their LDL had remained normal? Instead, we have to rely on the comparisons above (which consistently suggest that normal-LDL folks see much slower progression). 

Still, the point stands: this was effectively an observational cohort study. Without randomization or a control, confounding factors or regression to the mean could theoretically play some role. 

The authors did try to mitigate this by selecting a very homogeneous group and doing pre/post comparisons within the same individuals. But given that the entire cohort was exposed to high LDL, the study couldn’t directly test the effect of that exposure versus not being exposed. In essence, it told us what happened, but not definitively why, since everyone in the study shared the key exposure (very high apoB). This limitation was widely acknowledged and is important when drawing conclusions about LDL’s role.

Authors’ Rebuttals and Responses

The study authors and supporters have not been silent in the face of these critiques – they’ve offered several rebuttals and clarifications:

“Heterogeneity” Within LMHRs: 

The authors stress that there was wide variability in plaque changes among participants. Some LMHRs had rapid plaque growth, but others showed minimal change. They argue that this heterogeneity means we should be cautious about lumping all LMHRs together. 

In fact, Dave Feldman (an author on the study and prominent LMHR proponent) noted that a follow-up paper is in the works to analyze why certain LMHRs progressed more than others. Possible factors could include subtle differences in diet, genetics, baseline plaque, or other biomarkers. The authors seem genuinely interested in identifying a subset of LMHRs who might be at lower risk versus those at higher risk, which could enable personalized guidance in the future. 

Their emphasis is: not every LMHR had a bad outcome; some did fine, so what distinguishes them? This is a fair question and a useful line of inquiry – but as some critics responded, “let’s not miss the forest for the trees.” 

Yes, individuals vary, but the group average outcome was still clearly adverse. 

Highlighting heterogeneity can risk downplaying the overall trend (just as pointing out that some lifelong smokers live to 90 doesn’t refute that smoking is harmful on average). The rebuttal from the scientific community has been that while heterogeneity exists (as in any study), the burden of proof is now on showing that an LMHR can be low-risk, given that the majority were not. 

Until we know how to predict which LMHR might avoid plaque buildup, one must assume any LMHR could be susceptible. In summary, exploring variability is welcome – but it should not detract from the clear message that most of these people saw concerning plaque increases.

“Plaque Begets Plaque” – Not LDL (within this cohort): 

The central claim of the publication – as signaled by its title – is that baseline plaque burden was a much stronger predictor of progression than LDL levels. 

In the data, those with higher initial CAC, NCPV, or PAV had greater increases in NCPV over the year. Meanwhile, there was no significant correlation between individuals’ LDL or ApoB levels (which were all high, but ranged from ~150 to 300+ mg/dL LDL) and the amount of plaque growth. 

They interpret this as evidence that simply having very high LDL per se may not distinguish who gets disease – instead “something about having plaque already” predisposes further plaque. In press releases, Dr. Budoff and colleagues argued that “high cholesterol is not always a marker of plaque progression” in metabolically healthy people, and that LMHRs might benefit from direct imaging to assess risk rather than assumptions based on LDL. 

They also cite their previous cross-sectional finding that LMHRs had similar plaque as controls at baseline, suggesting that LDL elevation in these folks “may not indicate a higher risk of coronary plaque”. In essence, the authors’ stance is that traditional risk markers might not fully apply to this phenotype, and one should look at the arteries (plaque presence) to gauge risk.

Rebuttal to “No LDL correlation” – The Other Side: 

Many lipid experts responded that the lack of LDL-Plaque correlation within the homogeneous LMHR group is not surprising – and certainly does not mean LDL isn’t causing the plaque. 

It’s a classic epidemiological concept: range restriction. All the LMHRs had very high LDL (median ~237, nearly all well above 190), so there was no “low LDL” group to contrast against. 

It’s analogous to studying only heavy smokers: if everyone smokes 2–4 packs a day, you might find that cigarettes per day don’t correlate with lung disease severity – but that’s because the whole group is at the extreme end of exposure. 

You haven’t included any non-smokers for comparison. 

Similarly, in KETO-CTA, saying “ApoB does not predict plaque progression” is narrowly true for differences within this high-ApoB cohort, but it cannot answer whether the high ApoB exposure as a whole was driving the plaque increases. 

In fact, the authors themselves acknowledge the study “cannot address the broader question of whether high ApoB drives atherosclerosis” due to the lack of a low-ApoB control. The alternative interpretation is that all these LMHRs were at high risk from high LDL, and whether one gained 5 mm³ vs. 50 mm³ of plaque had more to do with how much plaque they’d already developed pre-study (or other genetic factors) than their exact LDL level (all of which were in the danger zone).

Baseline plaque predicting future plaque is not a novel concept – it’s been seen in other studies too. It likely indicates that once the disease process is initiated (even minimally), it can feed forward. 

Meanwhile, those who truly had zero plaque might simply need more time or had slightly protective factors delaying the onset. The key point from critics: do not misinterpret “no gradient of risk within a high-LDL group” as meaning “LDL is harmless.” 

A more appropriate conclusion is that virtually everyone in this LMHR cohort had sufficient LDL to promote atherosclerosis, and other factors determined who manifested it fastest. 

The data actually reinforce concern about LDL: despite pristine metabolic health, the group as a whole did poorly – which is exactly what one would expect if apoB is a dominant causal driver that can override other favorable traits. 

In short, the authors’ rebuttals center on: “some did fine, we need to find out why” and “LDL might not be telling us everything, look at baseline plaque.” The scientific responses have been: “heterogeneity or not, the average outcome is concerning” and “the lack of LDL correlation is a function of study design – it doesn’t exonerate LDL.” 

Overall, the clinical message is caution. Until proven otherwise, an LDL-C ≥200 mg/dL (or ApoB ~ ApoB ≥150–180 mg/dL) should be taken seriously even in otherwise metabolically healthy patients. The notion of a “warranted variation” for LMHRs (i.e. that their LDL doesn’t carry risk) is not supported by empirical evidence – quite the opposite, this study suggests risk is present and significant.

Final Thoughts

In conclusion, the KETO-CTA study underscores that extremely high LDL-C is a significant cardiovascular risk factor, even in the context of a lean, low-inflammation phenotype. 

It does not mean that keto diets or LMHRs are invariably headed for heart attacks – individual outcomes will vary and longer follow-up is needed to quantify absolute risk – but it soundly rejects the idea that the LMHR profile is protective or benign. 

The data argue that clinicians and patients should take LDL elevations seriously and weigh the risks, rather than assuming that “my HDL is high and CRP low, so I’m fine.” 

For now, this study offers an invaluable evidence-based perspective: when it comes to atherosclerosis, LDL still matters – a lot – even for those who appear metabolically superb.