JACC: ADVANCES VOL. -, NO.

-, 2025
ª 2025 THE AUTHORS. PUBLISHED BY ELSEVIER ON BEHALF OF THE AMERICAN

COLLEGE OF CARDIOLOGY FOUNDATION. THIS IS AN OPEN ACCESS ARTICLE UNDER

THE CC BY LICENSE (http://creativecommons.org/licenses/by/4.0/).

ORIGINAL RESEARCH

Plaque Begets Plaque, ApoB Does Not

Longitudinal Data From the KETO-CTA Trial

Adrian Soto-Mota, MD, PHD,a,b,* Nicholas G. Norwitz, PHD,c,* Venkat S. Manubolu, MD,d April Kinninger, MPH,d
Thomas R. Wood, BM BCH, PHD,e James Earls, MD,f David Feldman, AA,g,z Matthew Budoff, MDd,z

ABSTRACT

BACKGROUND Changes in low-density lipoprotein cholesterol (LDL-C) among people following a ketogenic diet (KD)
are heterogeneous. Prior work has identified an inverse association between body mass index and change in LDL-C.
However, the cardiovascular disease risk implications of these lipid changes remain unknown.

OBJECTIVES The aim of the study was to examine the association between plaque progression and its predicting
factors.

METHODS One hundred individuals exhibiting KD-induced LDL-C $190 mg/dL, high-density lipoprotein
cholesterol $60 mg/dL, and triglycerides #80 mg/dL were followed for 1 year using coronary artery calcium and cor-
onary computed tomography angiography. Plaque progression predictors were assessed with linear regression and Bayes
factors. Diet adherence and baseline cardiovascular disease risk sensitivity analyses were performed.

RESULTS High apolipoprotein B (ApoB) (median 178 mg/dL, Q1-Q3: 149-214 mg/dL) and LDL-C (median 237 mg/dL,
Q1-Q3: 202-308 mg/dL) with low total plaque score (TPS) (median 0, Q1-Q3: 0-2.25) were observed at baseline. Neither
change in ApoB (median 3 mg/dL, Q1-Q3: 17 to 35), baseline ApoB, nor total LDL-C exposure (median 1,302 days,
Q1-Q3: 984-1,754 days) were associated with the change in noncalcified plaque volume (NCPV) or TPS. Bayesian infer-
ence calculations were between 6 and 10 times more supportive of the null hypothesis (no association between ApoB and
plaque progression) than of the alternative hypothesis. All baseline plaque metrics (coronary artery calcium, NCPV, total
plaque score, and percent atheroma volume) were strongly associated with the change in NCPV.

CONCLUSIONS In lean metabolically healthy people on KD, neither total exposure nor changes in baseline levels
of ApoB and LDL-C were associated with changes in plaque. Conversely, baseline plaque was associated with plaque
progression, supporting the notion that, in this population, plaque begets plaque but ApoB does not. (Diet-induced
Elevations in LDL-C and Progression of Atherosclerosis [Keto-CTA]; NCT05733325) (JACC Adv. 2025;-:101686)
© 2025 The Authors. Published by Elsevier on behalf of the American College of Cardiology Foundation. This is an open
access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

From the aMetabolic Diseases Research Unit, National Institute for Medical Sciences and Nutrition Salvador Zubiran, Mexico City,
Mexico; bTecnologico de Monterrey, School of Medicine, Mexico City, Mexico; cHarvard Medical School Boston, Massachusetts,
USA; dLundquist Institute at Harbor-UCLA Medical Center, Torrance, California, USA; eDepartment of Pediatrics, University of
Washington School of Medicine, Washington, USA; fCleerly, New York, New York, USA; and the gCitizen Science Foundation,
Las Vegas, Nevada, USA. *These authors contributed equally to this work. zThese authors are the senior authors.
The authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’
institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information,
visit the Author Center.

Manuscript received December 13, 2024; revised manuscript received February 24, 2025, accepted February 24, 2025.

ISSN 2772-963X https://doi.org/10.1016/j.jacadv.2025.101686

2 Soto-Mota et al JACC: ADVANCES, VOL. -, NO. -, 2025
Plaque Begets Plaque: KETO-CTA - 2025:101686

C
ABBREVIATIONS arbohydrate-restricted diets (CRDs), and near-LMHR phenotypes remains an outstanding
AND ACRONYMS including very low carbohydrate question, with no prospective longitudinal data
ketogenic diets (KDs), are published on this unique population to date (please
ApoB = apolipoprotein B
increasing in popularity beyond the manage- refer to our prior editorial on conservative clinical
BMI = body mass index
ment of obesity and diabetes. At present, management).16
bHB = b-hydroxybutyrate
CRDs are being implemented for mental In this KETO-CTA study, we followed 100 persons
CAC = coronary artery calcium health disorders, 1-4 epilepsy, neurodegener- with LMHR or near LMHR phenotypes over 1 year
CCTA = coronary computed ative diseases,5,6 polycystic kidney disease,7 using high-resolution coronary computed tomogra-
tomography angiography
and autoimmune and inflammatory condi- phy angiography (CCTA) with artificial intelligence-
CRD = carbohydrate-restricted
tions,8-10 to name a few examples. While guided plaque quantification to characterize plaque
diet
some of these clinical use cases are backed progression in this phenotype along with factors that
HDL-C = high-density
lipoprotein cholesterol
by human randomized trials, many remain might predict plaque progression. Previously, we had
experimental. Nevertheless, there is a clear reported that, upon baseline scans, study participants
KD = ketogenic diet
and rising interest in the therapeutic appli- who had been on KD with hypercholesterolemia for a
LDL-C = low-density
lipoprotein cholesterol cations of CRD and KD. mean of 4.7 years exhibited no greater total plaque
LMHR = lean mass hyper-
An obstacle to the broad clinical imple- score (TPS) compared to a matched control group.17
responder mentation of CRD and KD are lipid changes
METHODS
NCPV = noncalcified plaque that occur in a minority of patients upon
volume carbohydrate restriction, characterized by
DESIGN AND STUDY POPULATION. This study was
PAV = percent atheroma large increases in low-density lipoprotein
volume preregistered at Clinical Trials.gov with the code
cholesterol (LDL-C) and associated apolipo-
TPS = total plaque score
(NCT05733325) and was conducted in compliance
protein B (ApoB). While there are manifold
with the principles of the Declaration of Helsinki and
factors contributing to increases in LDL-C and ApoB
relevant local laws and regulations.
on CRDs, “leanness” has been shown to be a major
Participants were recruited through social media.
source of heterogeneity. 11,12 For example, in a recent
Interested individuals were then prescreened based
set of meta-analyses of 41 human randomized
on the full eligibility criteria and the availability of
controlled trials,12 only those with participants with a
supporting medical documentation to confirm serum
mean body mass index (BMI) of <25 kg/m 2 as a group
markers.
exhibited increases in LDL-C on CRD; those with par-
The inclusion criteria were
ticipants having overweight of class I obesity exhibited
no change in LDL-C; and those with participants hav-  Being on a KD for $24 months
ing class II obesity exhibited decreases in LDL-C.  LDL-C #160 mg/dL from the last lipid panel drawn
Correspondingly, this meta-analysis also found an in- prior to adopting a KD
verse association between BMI and LDL-C change; and  LDL-C $190 mg/dL on the most recent laboratory
it also found that having a BMI of <25 kg/m 2 was >5 on a KD
times as powerful as being in the top quartile of satu-  An increase of $50% in LDL-C after adopting a KD
rated fat intake for predicting LDL-C change.  HDL-C $60 mg/dL
These other studies and observations have given  Triglycerides #80 mg/dL
rise to the characterization of the lean mass hyper-  Glycated hemoglobin <6.0%
responder (LMHR), 11 defined by a triad of elevated  Fasting glucose <110 mg/dL
LDL-C, along with elevated high-density lipoprotein  High-sensitivity C-reactive protein <2 mg/L
cholesterol (HDL-C) and low triglycerides that tends to
Exclusion criteria were
occur in lean, generally healthy individuals adopting a
CRD. The phenotype may potentially be explained by  Elevated blood pressure (systolic >130 mm Hg,
the lipid energy model, which is beyond the scope of diastolic >80 mm Hg)
this report but described in Norwitz et al.13  Type 2 diabetes or any lifetime use of antidiabetic
Irrespective of the metabolic and mechanistic medication
drivers of this LMHR phenotype, there is under-  Untreated hypothyroidism (thyroid stimulating
standable clinical concern, especially among those hormone >10 mIU/mL)
LMHRs with the most extreme profiles including LDL-  Renal insufficiency (calculated creatinine clear-
C levels in excess of 500 mg/dL with correspondingly ance of <50 mL/min with the MDRD [Modification
high ApoB levels. 14,15 However, while clinicians of Diet in Renal Disease Study] equation)
should exercise clinical prudence in their individual  Liver enzymes >2 times the upper limit of normal
practices, the absolute risk associated with the LMHR at screening visit or total bilirubin >1.5
JACC: ADVANCES, VOL. -, NO. -, 2025 Soto-Mota et al 3
- 2025:101686 Plaque Begets Plaque: KETO-CTA

 Use of medications that elevate LDL-C (anabolic analyses, and all plots were made using R version
steroids, isotretinoin, immunosuppressant, amio- 4.0.3 (R Core Team [2020], R Foundation for Statis-
darone, thiazide diuretics, glucocorticoids, or tical Computing, with the last available version for
thiazolidinediones) September 2024). Cardiovascular risk was calculated
 Use of lipid-lowering supplements or medications with CVrisk::chd_10y_mesa. Linear models on the
(statins, red yeast rice, garlic, ezetimibe, berberine, primary (NCPV) and secondary outcomes were uni-
PCSK9 inhibitors) variable and analyzed using stats::lm, and all linear
 Genetically defined familial hypercholesterolemia model assumptions were corroborated with the R
function performance::check_model.
MEASUREMENTS. All participants were asked to stay Bayes factors were calculated using BayesFactor::-
on a KD during their follow-up, and to measure regressionBF with default settings and an w rscale
adherence, 3 dietary recalls and daily b-hydrox- value of 0.8 to contrast a moderately informative
ybutyrate ( b HB) data were collected using the Auto- prior with a conservative distribution width (to allow
mated Self-Administered 24-Hour (ASA24) Dietary for potential large effect sizes) due to the well-
Assessment Tool from the National Institutes of documented association between ApoB changes and
Health. Blood b HB monitoring devices by KetoMojo coronary plaque changes. 20
(Napa) were provided to each participant. Sensitivity analyses were performed on partici-
All CCTA scans were performed at baseline and pants with high diet adherence (defined as 80% bHB
1 year after at The Lundquist Institute for Biomedical measurements $0.3) and participants with a 10-year
Innovation at Harbor-UCLA Medical Center using the baseline cardiovascular risk >5% assessed with
256-multidetector computed tomography (GE Revo- the MESA (Multi-Ethnic Study of Atherosclerosis)
lution, General Electric) and were blindly read by the equation with ethnicity and CAC inputs (Central
level 3 cardiac computed tomography readers. A Illustration).
nonenhanced electrocardiogram-gated coronary ar-
tery calcium (CAC) scan was also performed before RESULTS
each CCTA.
Plaque volume quantification was measured using All 100 participants completed baseline and follow-up
a semiautomated software, Cleerly, and evaluated by measurements (Supplemental Figure 1). As is charac-
experienced readers. This software can detect lumen teristic of the LMHR phenotype, mean LDL-C, HDL-C,
and vessel border contours automatically, with and triglycerides were 254  85 mg/dL, 89  20 mg/dL,
manual correction by expert readers in any areas of and 67  30 mg/dL, respectively, and mean BMI
misregistration. Each coronary plaque area identified was 22.5  2.7 kg/m 2. Mean age was 55.3  10.7 years,
in at least 2 adjacent slides with a 0.6 mm slice and 59% were males (Table 1). While most parti-
thickness that was assessed by evaluating all affected cipants exhibited a TPS of 0 at baseline as assessed by
slides. Plaque volume was then calculated through blinded expert inspection (as previously reported17),
the multiplication of the area by slice thickness. The application of modern artificial intelligence-guided
summation of the luminal diameter and segments CCTA assessment was able to identify quantifiable
was calculated and is reported as “noncalcified,” “low plaque in all participants at baseline, as expected.
attenuation,” or “calcified.” This quantitative plaque Overall participants successfully adhered to their
assessment protocol has been widely used in several KD as confirmed by detailed dietary assessment
studies. 18 Further methodological details can be (Table 2) and capillary b HB measurements, with 87%
found in the published protocol for this study.19 of the cohort recording values above $0.3 mmol/L on
LDL-C exposure on a KD was calculated by summing a majority of measurements, a threshold chosen
the products of the reported days on a KD prior to study given the known steady-state drop in circulating ke-
commencement and baseline LDL-C on a KD plus the tone levels with prolonged adaptation.
study follow-up days by their final LDL-C. Estimated No significant changes in either ApoB and BMI
lifelong LDL-C additionally included the product of age were observed after 1 year (median 3 mg/dL, Q1-
upon commencing a KD and pre-KD LDL-C. The Q3: 17 to 35) and (median 0.2, Q1-Q3: 0.7 to 0.2),
percent atheroma volume (PAV) was calculated as respectively.
(total plaque volume/total vessel volume)  100. Most participants presented with stable NCPV
STATISTICAL ANALYSES. Data are presented using (Figures 1A and 1B), with 1 participant exhibiting a
the median (25th to 75th percentiles) and mean  SD decrease in NCPV (Figures 2A to 2C) and 6 participants
for continuous variables and count (percentage) for showing decreases in TPS scores over 1 year
categorical variables. Data cleaning, statistical (Figures 2D to 2F). The median change in PAV was
4 Soto-Mota et al JACC: ADVANCES, VOL. -, NO. -, 2025
Plaque Begets Plaque: KETO-CTA - 2025:101686

C E NT R A L IL L U ST R A T I O N 100 Individuals Exhibiting Ketogenic Diet-Induced Low-Density Lipoprotein Cholesterol

Elevations Were Followed for 1 Year Using Coronary Artery Calcium and Coronary Computed Tomography Angiography

Method
n HDL-C nLDL-C

pTG
KETO LDL-C: 254 ± 51 mg/dL
n = 100 ApoB: 185 ±51 mg/dL
Plaque Progression (CCTA)
HDL-C: 89 ±20 mg/dL
TG: 67 ±30 mg/dL

Results
Change in Total Plaque Score

Over a 1-year prospective trial

conducted on lean, metabolically healthy
Change in NCPV (mm3)

persons with carbohydrate-restriction

induced elevations in ApoB, ApoB did not
predict plaque accumulation. Baseline CAC
and other plaque metrics did predict
plaque accumulation, suggesting plaque
begets plaque whereas ApoB is not a
significant predictor of risk in this
population.

ApoB (mg/dL) Baseline CAC Score

Soto-Mota A, et al. JACC Adv. 2025;-(-):101686.

Neither total exposure nor changes in baseline levels of ApoB and mg/dL were associated with changes in plaque. Conversely, baseline plaque but ApoB
was not associated with plaque progression. ApoB ¼ apolipoprotein B; CAC ¼ coronary artery calcium; CCTA ¼ coronary computed tomography angiography;
HDL-C ¼ high-density lipoprotein cholesterol; LDL-C ¼ low-density lipoprotein cholesterol; TG ¼ triglycerides.

0.8%, and both NCPV and PAV values were compa- TPS (Figure 2G, Table 3). Estimated lifetime LDL-C
rable with those observed in other cohorts on both exposure was only a significant predictor of final
visits (Figures 1A and 1B, Supplemental Table 1). NCPV in the univariable analysis but lost signifi-
Neither the change in ApoB throughout the study cance when age was included as a covariate
nor the ApoB level on a KD at the beginning of the (Table 3). Both age and lifetime LDL-C exposure lost
study were associated with the change in NCPV significance when baseline CAC was included in the
(Figures 2A and 2B, Table 3) nor with TPS (Figures 2D model (Table 3).
and 2E, Table 3). All associations provided Bayes factors 6 to 10
By contrast, baseline CAC was positively associated times more likely under the null (for change in and
with a change in NCPV (Figure 2C) ( b ¼ 1.8, P < 0.001, baseline ApoB vs changes in NCPV) and 10 times more
R 2 ¼ 0.33) (Figure 2F). Similar results were found for likely under the alternative hypotheses (for baseline
other baseline plaque metrics, where baseline NCPV, plaque metrics [CAC, NCPV, TPS, and PAV] vs changes
TPS, and PAV were positively associated with change in NCPV) (Table 3).
in NCPV (Table 3). Sensitivity analyses on participants with >80% of
There was no association between LDL-C expo- b HB measurements above 0.3 mmol/L (Supplemental
sure while on a KD (mean 5.7 years) and NCPV or Tables 2 to 4) and with high calculated 10-year
JACC: ADVANCES, VOL. -, NO. -, 2025 Soto-Mota et al 5
- 2025:101686 Plaque Begets Plaque: KETO-CTA

cardiovascular risk showed similar results to those

T A B L E 1 Baseline Characteristics (N ¼ 100)
just reported (Supplemental Table 5).
Male 59%
Ethnicity
DISCUSSION
Asian 10%
Hispanic 7%
While both LDL-C and ApoB are independent risk White 83%
factors for atherosclerosis, the absolute risk associ- Age (y) 55.3  10.6 57.0 (50.8-63.0)
ated with elevated LDL-C and ApoB is context- Ketogenic diet duration (d) 1,642.7  913.5 1,427 (1,002-1,938)

dependent, including the etiology of the elevations 21 Body mass index (kg/m2) 22.5  2.7 22.3 (20.6-24.4)
Systolic blood pressure (mm Hg) 121.8  21.1 120 (112-135)
in these biomarkers as well as interactions with
22 Total cholesterol (mg/dL) 355.1  89.9 338 (301-337)
other risk markers. Thus, these data are consistent
LDL-C (mg/dL) 253.7  84.7 237 (202-308)
with the observation that high LDL-C and ApoB
HDL-C (mg/dL) 88.8  20.0 88 (74-102)
among a metabolically healthy population have Triglycerides (mg/dL) 66.7  30.4 61 (52-77)
different cardiovascular risk implications than Apolipoprotein B (mg/dL) 185.0  50.8 178 (149-214)
high LDL-C among those with metabolic dysfunction, Noncalcified plaque volume (mm3) 75.9  88.3 44 (15.4-102.3)
who constitute a majority proportion of the Coronary artery calcium score 50.3  100.9 0 (0-54)

population. 23 Percent atheroma volume 3.2  3.8 1.25 (0.5-3.6)

Total plaque score 1.7  2.6 0 (0-2.0)
It is worth emphasizing that the population at
10-y CVD risk (MESAETH-CAC, %) 5.2  5.3 3.4 (2.0-5.9)
hand—LMHR and near-LMHR persons on KD—is
distinct from any population described previously in Values are %, mean  SD, or median (Q1-Q3). 10-y CVD risk ¼ based on the risk equation from the
MESA (Multi-Ethnic Study of Atherosclerosis) with ethnicity and CAC inputs.
several ways.
CAC ¼ coronary artery calcium; CVD ¼ cardiovascular disease; HDL-C ¼ high-density lipoprotein
cholesterol; LDL-C ¼ low-density lipoprotein cholesterol.
(1) Elevations in LDL-C and ApoB are dynamic and
result from a metabolic response to carbohydrate
restriction,13 rather than as a function of a
congenital defect. manner in this population of metabolically healthy
(2) These participants are of normal healthy weight individuals with carbohydrate restriction-induced
(BMI <25 kg/m 2) and metabolically healthy, rather elevations in LDL-C and ApoB.
than living with obesity, prediabetes, type 2 dia- However, this does not mean that the LMHR pop-
betes, or other insulin resistance disorders. 11,14 ulation is without risk. The observed, the observed
(3) The high LDL-C and ApoB in this phenotype PAV progression in this KETO-CTA cohort was com-
emerge as part of a lipid triad, also inclusive of parable to what has been observed in other studies on
high HDL-C and low triglycerides, representing a populations with lower LDL-C across the cardiovas-
metabolic signature of a distinct physiological cular disease risk spectrum. It should be emphasized
state. that this includes heterogeneity in progression (and
(4) The degree of this phenotype appears inversely regression) across the population.
related to BMI (“leanness”), 12 consistent with the However, the question that should follow is: “What
idea that it is a metabolic response to carbohy- explains this heterogeneity?” Here, ApoB exposure was
drate restriction that is accentuated in leaner, not a significant predictor of changes in plaque,
more metabolically healthy persons. whereas all baseline plaque metrics were significant
predictors of changes in plaque.
These points are worth emphasizing because
At this time, we can only speculate about the lack
unique populations require independent study to
of association between ApoB and plaque progression
properly characterize the risk associated with their
metabolic profiles. Furthermore, this is the first and
only population in which LDL-C is independently
elevated without any clear underlying congenital T A B L E 2 Diet Composition (N ¼ 100)
genetic cause and outside the context of any other Carbohydrates (g/d) 39.5  29.8 30.8 (12.6-51.1)
notable metabolic dysfunction. Fat (g/d) 120.2  46.4 113.1 (89.8-139.3)
Thus, the LMHR population constitutes a unique Saturated fat (g/d) 44.7  18.9 39.4 (29.9-52.6)
and important natural experiment evaluating the Sugar (g/d) 20.2  18.5 11.1 (6.3-22.2)

lipid heart hypothesis in an unprecedented manner. 16 Fiber (g/d) 6.9  6.5 6.9 (1.3-11.3)
Sodium (mg/d) 3,292.5  1,263.0 2,931.4 (2,217.7-3,265.7)
The data presented herein are consistent with the
notion that elevated ApoB, even at extreme levels, Values are mean  or median (Q1-Q3). Pooled data from all available dietary records.
does not drive atherosclerosis in a dose-dependent
6 Soto-Mota et al JACC: ADVANCES, VOL. -, NO. -, 2025
Plaque Begets Plaque: KETO-CTA - 2025:101686

F I G U R E 1 Individual Change in Plaque Volume

N ¼ 100 individual-level changes in noncalcified plaque volume (NCPV) as measured by artificial intelligence-guided quantification of NCPV at
baseline and 1-year follow-up (A). The red line represents the median, and the shaded area represents the IQR.

and mechanisms driving the progression in this pop- STRENGTHS AND LIMITATIONS. A major strength of
ulation. Plausible explanations include diet- and this study lies in the nature of the participants, who
health-related differences in lipoprotein quality 24 collectively constitute a population of 100 metabolic
and kinetics, insulin sensitivity, or inflammation,25 “outliers,” all characterized by high LDL-C and
as compared to the general population. Future ApoB in the context of being metabolically healthy
studies will have to directly assess and compare these and generally lean with drastic changes in LDL-C
potential mechanisms. brought about by carbohydrate restriction. In
These insights can facilitate personalized treat- addition, the dropout rate was 0%, and adherence
ment and risk mitigation strategies based on mod- to the KD was assessed by dietary records and
ern, cost-effective cardiac imaging. For instance, b HB measurements.
despite profound elevations in LDL-C and ApoB, Since lack of statistical significance (ie, P > 0.05)
based on these data, LMHR subjects with CAC ¼ 0 should not be interpreted as evidence in favor of
at baseline (n ¼ 57) constitute a low-risk group for the null but simply a failure to reject the null, the
PAV progression, even as compared to other cohorts addition of Bayesian inference adds credence to
with far lower LDL-C and ApoB. By contrast, LMHR finding that there is no association between NCPV
subjects with elevated baseline CAC, possibly from a vs LDL-C or ApoB and TPS vs LDL-C or ApoB. In
history of metabolic damage and dysfunction prior other words, these data suggest it is 6 to 10 times
to adopting a CRD, appear to constitute a relatively more likely that the hypothesis of no association
higher risk group for PAV progression even where between these variables (the null) is true as
LDL-C and ApoB are equal to their CAC ¼ 0 compared to the alternative.
counterparts. Of course, this study is not without limitations.
The finding that “plaque begets plaque” could The duration of the prospective trial was 1 year.
result from underlying susceptibilities in those with However, given the quality of modern CCTA imaging,
greater amounts of pre-existing plaque or could it is generally agreed that this is sufficient to observe
result, more directly, from the proinflammatory state and quantify changes in NCPV, as we did. Nonethe-
associated with coronary atheroma. More research less, we should mention that the “normal” values for
will be needed to identify the pathophysiology NCPV, PAV, and plaque progression in a healthy
driving the progression of atherosclerosis in people population are yet to be fully determined, as there is a
with higher levels of pre-existing plaque. wide range of ever-evolving methods, definitions,
JACC: ADVANCES, VOL. -, NO. -, 2025 Soto-Mota et al 7
- 2025:101686 Plaque Begets Plaque: KETO-CTA

F I G U R E 2 Changes in Noncalcified Plaque Volume and Total Plaque Score vs Apolipoprotein B and Coronary Artery Calcium

(A to C) Change in ApoB, baseline ApoB, and baseline CAC vs NCPV. (D to F) Change in ApoB, baseline ApoB, and baseline CAC vs NCPV. (G) Final NCPV vs diet LDL-C
exposure. (H) ApoB vs saturated fat intake. (C, F) Only CAC is associated with changes in NCPV and TPS. The regression line was fitted with the function “lm,” which
regresses ywx, and the shaded area represents the standard error. ApoB ¼ apolipoprotein B; CAC ¼ coronary artery calcium; LDL-C ¼ low-density lipoprotein
cholesterol; NCPV ¼ noncalcified plaque volume; TPS ¼ total plaque score.
8 Soto-Mota et al JACC: ADVANCES, VOL. -, NO. -, 2025
Plaque Begets Plaque: KETO-CTA - 2025:101686

T A B L E 3 Model Results

b P Value R2 BF

Estimated LDL-C exposure

NCPVfinal w LDL-Cexp 0.00 0.88 0.01 (01) 9.9
PAVfinal w LDL-Cexp 0.00 0.73 0.01 (01) 9.1
ApoB
DNCPV w DApoB 0.01 0.91 0.01 (01) >10.0
DNCPV w ApoB 0.06 0.33 0.00 (01) 6.3
Plaque Metrics
DNCPV w CACbl 0.18 <0.001 0.33 (10) >10.0a
DNCPV w NCPVbl 0.25 <0.001 0.49 (10) >10.0a
DNCPV w PAVbl 5.5 <0.001 0.43 (10) >10.0a
DNCPV w TPSbl 7.4 <0.001 0.37 (10) >10.0a
DNCPV w CACbl* DApoB
CACbl 0.18 <0.001 0.33 N/A
DApoB 0.08 0.20 N/A N/A
CACbl: DApoB 0.00 0.23 N/A N/A
Saturated fat
DNCPV w Saturated fat intake 0.05 0.77 0.01 (01) 9.8
ApoB w Saturated fat intake 0.03 0.90 0.01 (01) >10.0
Age mediation analysis
NCPVfinal w Age 3.0 0.004 0.07 (01) 0.3
NCPVfinal w Age þ Life-LDL-Cexp 1.93 | 0.01 0.13 | 0.16 0.07 (01) 0.3 | 0.3
NCPVfinal w Age þ Life-LDL-Cexp þ CACbl 0.02 | 0.01 | 0.73 0.98 | 0.15 | <0.001 0.47 (01) 2.7 | 3.1 | >10.0a

a
Models on CAC are provided for the alternative hypothesis (10). All other models are provided for the null hypothesis (01).
b ¼ estimate (slope magnitude); DNCPV ¼ change in noncalcified plaque volume; ApoB ¼ apolipoprotein B; ApoB ¼ ApoB on a ketogenic diet; DApoB ¼ ApoB change during
the study; BF ¼ Bayes factor; CACbl ¼ CAC at baseline; LDL-Cexp ¼ LDL-C exposure while on a ketogenic diet (mean 5.7 y); Life-LDL-Cexp ¼ LDL-C exposure over life course to
date; NCPVfinal ¼ noncalcified plaque volume at the end of the study; PAV ¼ percent atheroma volume; R2 ¼ squared correlation coefficient (explained variability);
TPSfinal ¼ total plaque score at the end of the study; other abbreviations as in Table 1.

and analytic techniques. We also do not have a FUNDING SUPPORT AND AUTHOR DISCLOSURES
comparator group; however, a carefully matched
comparison of 80% of this cohort showed similar and This study was funded by the Citizen Science Foundation, 7,320 S
Rainbow Blvd, #102 to 182, Las Vegas, NV, United States. Dr Nor-
trending lower levels of coronary plaque as compared
witz is coauthor of a Mediterranean low-carbohydrate-diet cook-
to other generally healthy populations with lower book; and he donates all royalty payments to nutrition research
LDL-C, as reported previously. 17 and education. Dr Feldman has received financial contributions
from membership (eg, through Patreon) for continued research and
is a partner in Own Your Labs LLC. All other authors have reported
CONCLUSIONS
that they have no relationships relevant to the contents of this
paper to disclose.
Over a 1-year prospective study of 100 persons
exhibiting extreme carbohydrate restriction-induced ADDRESS FOR CORRESPONDENCE: Dr Adrian Soto-
elevations in LDL-C and ApoB, changes in and base- Mota, Metabolic Diseases Research Unit, National
line levels of ApoB were not associated with changes Institute for Medical Sciences and Nutrition Salvador
in noncalcified plaque volume or total plaque score as Zubiran, Mexico City, Mexico. E-mail: adrian.sotom@
measured by CCTA. However, baseline CAC and other incmnsz.mx. OR Dr Nicholas G. Norwitz, Harvard
plaque metrics were positively associated with in- Medical School Boston, Massachusetts. E-mail:
creases in coronary plaque, supporting the notion nicholas_norwitz@hms.harvard.edu. OR Dr Matthew
that plaque begets plaque but ApoB does not, in this Budoff, UCLA, Endowed Chair of Preventive Cardiol-
population. ogy, Lundquist Institute, 1124 W Carson Street, Tor-
ACKNOWLEDGMENTS The authors thank the partic- rance, California 90502, USA. E-mail: mbudoff@
ipants who made this research possible. lundquist.org.
JACC: ADVANCES, VOL. -, NO. -, 2025 Soto-Mota et al 9
- 2025:101686 Plaque Begets Plaque: KETO-CTA

PERSPECTIVES

COMPETENCY IN MEDICAL KNOWLEDGE AND- TRANSLATIONAL OUTLOOK: Understanding the true

COMMUNICATION SKILLS: LMHR is an emerging risk associated with the LMHR phenotype, and high LDL-C
phenotype of growing research interest, with little known and ApoB among lean, metabolically healthy persons on
previously with respect to mechanism and risk. Physi- ketogenic diets, will take an open-minded multidisci-
cians’ awareness of unique aspects of the phenotype and plinary approach. In the opinion of this team, research
unique cardiovascular risk profile can facilitate individu- priorities should include dissecting the driving mechanism
alized patient management and future collaborative behind the LMHR phenotype by testing elements of the
research effects. Furthermore, patients presenting with lipid energy model, along with comparing lifestyle and
the LMHR phenotype often identify with the phenotype pharmacological treat options in LMHR in randomized
and may be more receptive to clinical advice from phy- controlled trials. While these first prospective data should
sicians who acknowledge the unique aspects of their be reassuring for LMHR patients and their physicians,
profile, are aware of new and ongoing research in this longer-term follow-up (eg, 2-, 5-year CCTA measure-
area, and engage in open discussion of the knowns and ments) of this and similar cohorts will also be essential to
unknowns with these patients. confirm that LMHR and near-LMHR constitute, as a pop-
ulation, a low cardiovascular risk profile.

REFERENCES

1. Danan A, Westman EC, Saslow LR, et al. The 9. Ciaffi J, Mitselman D, Mancarella L, et al. The ef- hyper-responders on low-carbohydrate ketogenic
ketogenic diet for refractory mental illness: a fect of ketogenic diet on inflammatory arthritis and diets deserve urgent clinical attention and further
retrospective analysis of 31 inpatients. Front Psy- cardiovascular health in rheumatic conditions: a mini research. J Clin Lipidol. 2023;16(6):765–768.
chiatry. 2022;13:951376. review. Front Med (Lausanne). 2021;8:792846.
17. Budoff M, Manubolu VS, Kinninger A, et al.
2. Norwitz NG, Sethi S, Palmer CM, et al. Keto- 10. Lambadiari V, Katsimbri P, Kountouri A, et al. Carbohydrate restriction-induced elevations in
genic diet as a metabolic treatment for mental The effect of a ketogenic diet versus mediterra- LDL-cholesterol and atherosclerosis: the KETO
illness. Curr Opin Endocrinol Diabetes Obes. nean diet on clinical and biochemical markers of trial. JACC Adv. 2024;3(8):101109.
2020;27(5):269–274. inflammation in patients with obesity and psoriatic 18. Abd Alamir M, Ellenberg SS, Swerdloff RS,
arthritis: a randomized crossover trial. Int J Mol et al. The Cardiovascular Trial of the Testosterone
3. Sethi S, Wakeham D, Ketter T, et al. Ketogenic
Sci. 2024;25(5):2475. Trials: rationale, design, and baseline data of a
diet intervention on metabolic and psychiatric
health in bipolar and schizophrenia: a pilot trial. 11. Norwitz NG, Feldman D, Soto-Mota A, et al. clinical trial using computed tomographic imaging
Psychiatry Res. 2024;335:115866. Elevated LDL-cholesterol with a carbohydrate- to assess the progression of coronary atheroscle-
restricted diet: evidence for a ‘lean mass hyper- rosis. Coron Artery Dis. 2016;27(2):95–103.
4. Norwitz NG, Hurn M, Espi Forcen F, et al. Ani-
responder’ phenotype. Curr Dev Nutr. 2021;6(1): 19. Javier DAR, Manubolu VS, Norwitz NG, et al.
mal-based ketogenic diet puts severe anorexia
nzab144. The impact of carbohydrate restriction-induced
nervosa into multi-year remission: a case series. J
Insulin Resist. 2023;6:a84. 12. Soto-Mota A, Flores-Jurado Y, Norwitz NG, elevations in low-density lipoprotein cholesterol
et al. Increased LDL cholesterol in adults with on progression of coronary atherosclerosis: the
5. Phillips MCL, Deprez LM, Mortimer GMN, et al. ketogenic diet trial study design. Coron Artery Dis.
normal but not high body weight: a meta-analysis.
Randomized crossover trial of a modified keto- 2024;35(7):577–583.
Am J Clin Nutr. 2024;119(3):740–747.
genic diet in Alzheimer’s disease. Alzheimer’s Res
20. Shui X, Wen Z, Dong R, et al. Apolipoprotein B
Ther. 2021;13:51. 13. Norwitz NG, Soto-Mota A, Kaplan B, et al. The
is associated with CT-angiographic progression
lipid Energy model: reimagining lipoprotein func-
6. Vanitallie TB, Nonas C, Di Rocco A, et al. beyond low-density lipoprotein cholesterol and
tion in the context of carbohydrate-restricted di-
Treatment of Parkinson disease with diet-induced non-high-density lipoprotein cholesterol in pa-
ets. Metabolites. 2022;12(5):460.
hyperketonemia: a feasibility study. Neurology. tients with coronary artery disease. Lipids Health
2006;66(4):617. author reply 617. 14. Needham N, Campbell IH, Grossi H, et al. Pilot Dis. 2023;22(1):125.
study of a ketogenic diet in bipolar disorder.
7. Cukoski S, Lindemann CH, Arjune S, et al. 21. Trinder M, Francis GA, Brunham LR, et al. As-
BJPsych open. 2025;9(6):e176.
Feasibility and impact of ketogenic dietary in- sociation of monogenic vs polygenic hypercho-
terventions in polycystic kidney disease: KETO- 15. Norwitz NG, Soto-Mota A, Feldman D, et al. lesterolemia with risk of atherosclerotic
ADPKD-a randomized controlled trial. Cell Rep Case report: hypercholesterolemia "Lean Mass cardiovascular disease. JAMA Cardiol. 2020;5(4):
Med. 2023;12:101283. Hyper-Responder" phenotype presents in the 390–399.
context of a low saturated fat carbohydrate-
8. Norwitz NG, Soto-Mota A. Carnivore-ketogenic 22. Mortensen MB, Caínzos-Achirica M,
restricted diet. Front Endocrinol. 2022;13:830325.
diet for the treatment of inflammatory bowel Steffensen FH, et al. Association of coronary pla-
disease: a case series of 10 patients. Front Nutr. 16. Norwitz NG, Mindrum MR, Giral P, et al. que with low-density lipoprotein cholesterol
2024;11:1467475. Elevated LDL-cholesterol levels among lean mass levels and rates of cardiovascular disease events
10 Soto-Mota et al JACC: ADVANCES, VOL. -, NO. -, 2025
Plaque Begets Plaque: KETO-CTA - 2025:101686

among symptomatic adults. JAMA Netw Open. 25. Wu X, Roussell MA, Hill AM, et al. Baseline angiography, carbohydrate-restricted diet,
2022;5(2):e2148139. insulin resistance is a determinant of the small, ketogenic diet, LDL cholesterol, lean mass
dense low-density lipoprotein response to diets hyper-responder, saturated fat
23. Libby P. The changing landscape of athero-
differing in saturated fat, protein, and carbohy-
sclerosis. Nature. 2021;592(7855):524–533.
drate contents. Nutrients. 2021;13(12):432.
24. Dugani SB, Moorthy MV, Li C, et al. Associa-
tion of lipid, inflammatory, and metabolic bio-
markers with age at onset for incident coronary A PPE NDI X For supplemental tables and a
heart disease in women. JAMA Cardiol. 2021;6(4): KEY WORDS ApoB, atherosclerosis, body figure, please see the online version of this
437–447. mass index, coronary computed tomography paper.