KETO-CTA: Clarifications and Moving Forward

Well, if you've been following, it's certainly been an "interesting time." Now, it's time to break the silence on KETO-CTA. Let's chat...

Since our KETO-CTA paper was published on April 7, 2025 there has been an undeniable and conspicuous spiral of events, leading to a strong diverse set of opinions on the data. These data were always going to be controversial. The controversy around the study of lean mass hyper-responders has been marinating for years. That doesn’t mean we didn’t make mistakes, but it does explain why events exploded as they did.

It’s also been noted that since around April 18^th, my co-authors and I have been quiet regarding criticisms rendered. I’ll speak for myself when I say this wasn’t personal my preferred approach. (Read between the lines on that as you will.) However, it was the strong preference of JACC Advances that we work through the preferred academic channels – namely, by responding to Letters to the Editors passed to us from the journal.

Now that we’ve done so (Click HERE), I’m pleased to break my silence and speak more freely.

First and foremost, I encourage everyone to listen to this recent hour-long conversation between Dave Feldman and Chris MacAskill about the controversy. Truthfully, I think it was among the most honest, humble, and sincere conversations my ears have ever had the pleasure of capturing. Please start there if you’ve been following the controversy and want a grounding perspective.

The April 7th Paper was Fully Peer-Reviewed. But it Wasn’t Perfect.

Now, here are some points I’d like to make to begin to reanimate productive discussion and resolve confusions.

First, let’s get this straight, the April 7^th published paper was fully peer-reviewed and approved by all the authors. That’s not in question and was previously clarified in a joint statement.

Owning Our Mistake, But Operating Through Preferred Channels

Second, we made a mistake in not reporting the numerical values for NCPV progression. This was not an intentional omission, and the data were presented clearly and graphically in Figure 1, with a big, bold red line. It was not hard to approximate the numbers from the figure, clearly.

This was an oversight, and we offered to provide a corrigendum on the paper that included the NCPV relative and absolute numeric values. In fact, we advocated for (strongly) it but were declined on account of the fact that there were no “factual” errors in the final manuscript and, thus, a corrigendum would go against journal policy.

We were asked, instead, to provide the values in our response(s) to Letters to the Editor, which we not only did but also provided more transparency and data than were requested. (See Open Access Table Release below.)

This is why my newsletter has waited until this moment, so we could meet this obligation to the journal. Our reply to the letters is live, which you can visit HERE. And thankfully, we can now engage more on this important discussion with social media.

We have, and will continue to, invite constructive scrutiny. As for the non-constructive and personal attacks. Well, water off a duck’s back because – frankly – there are better ways to spend energy and brain power, including discussing the core questions and curiosities that these data have provoked.

What People Are Overlooking

Saturation Effect

One strength of our study is that it has the greatest spread of LDL-C (and ApoB), of any study ever conducted, if I’m not mistaken. Incredibly, we have LDL-C levels spanned >400 mg/dl range. Despite this, neither LDL-C nor ApoB showed any power in predicting plaque progression.

Some have countered that there may be a “saturation effect” because there was no low LDL-C metabolically healthy control group. In other words, “well, all the subjects had high LDL-C,” which is presumed to be “suboptimal.”

So, how do we know they didn’t just hit a ceiling?

Consider that the current lipid-heart hypothesis very clearly predicts a dose-response relationship such that more LDL/ApoB exposure is supposed to predict more plaque. All else being equal – that’s a central component of the lipid-heart hypothesis. So, if one is to postulate a saturation effect, they are – in effect – rewriting the present day lipid hypothesis as we know it. Indeed, to quote the famous EAS 2017 consensus, “…Mendelian randomization studies, and randomized intervention trials all demonstrate a remarkably consistent dose-dependent log-linear association between the absolute magnitude of exposure to LDL-C and the risk of ASCVD…”

Additionally, some have suggested out n = 100 study isn’t powered to detect an association between LDL/ApoB and plaque progression. To these folks I have three remarks:

(i) If sample sizes are your hangup, then should we just discard the Nobel Prize-winning work of Brown and Goldstein, who studied just a handful of patients with homozygous familial hypercholesterolemia, far fewer than were included in our study?

(ii) Given the enormous spread (>400 mg/dl) in this study, if the predictive power of LDL and ApoB is so tiny as to not be detectable in an n = 100 sample, what does that imply?

(iii) If you’d like to learn more about the statistics in this study and Bayes factors, Dr Soto Mota, MD PhD BSJ (Biostatistician Jedi) would love to give you a lesson.

It’s Still Not the ApoB

Building on this, I think it’s fair to claim that ApoB doesn’t predict plaque progression in this cohort. Honestly, I don’t think that’s a disputed finding.

That said, the statement “ApoB doesn’t predict plaque progression,” is distinct from, and consistent with, the statement “ApoB is causal in atherosclerosis.”

Something can be causal, and not predictive, like the Jenga block at the bottom of the tower that nobody will ever touch, but that is necessary to hold the tower in place.

With respect to the title “Plaque Begets Plaque, but ApoB does Not,” was meant to mirror the proverb “Money begets money” and to highlight the strong and clinically relevant association of baseline plaque values with plaque progression rate. In retrospect, “Plaque Predicts Plaque” would have been better to avoid misinterpretations. Point conceded and lesson learned, not all play on words land as they were intended.

What is “Stable Plaque?”

While there is no consensus on what “stable plaque” means for NCPV changes, we would like to emphasize that, since most participants showed low-baseline plaque values, relative changes are expected to be magnified. Furthermore, it’s true that most of the participants showed PAV changes below the agreed upon definition for rapid plaque progression of >1% per year.

Match Analysis with Miami Heart

Consider this: We conducted a match analysis to the Miami Heart population, including semi-quantitative and quantitative Total Plaque Score (TPS) and NCPV, respectively. This match was conducted 4.7 years, on average, after the KETO-CTA participants had started on a ketogenic diet and exhibited mean LDL-C levels of ~272 mg/dl.

At this 4.7-year mark, the two populations were comparable, with KETO-CTA subjects trending lower for TPS and both populations have NCPV ~47 mm3.

This being said, there are interesting findings we’re looking into deeper, particularly where the KETO-CTA cohort matched with MiHeart at the baseline, (KETO-CTA having been at elevated LDL-C levels for an average of 4.7 years) before the “pronounced increase” in the NCPV analysis for the year following (then an average of 5.7 years). This was also where the quantitative data was more weakly correlated with the semi-quantitative of our study, which we are actively exploring at this time, in an ongoing search for better understanding.

Open-Access Table: When Measures Diverge.

Arguably the most important announcement we have for you today is the release of our Open Access Table. For all 100 participants at both visits, it includes the Total Plaque Score (TPS) and Coronary Artery Calcification (CAC) scores from the semi-quantitative analysis, and the Non-Calcified Plaque Volume (NCPV), Total Calcified Plaque Volume (TCPV), and Percent Atheroma Volume (PAV) from the quantitative analysis. Visit https://citizensciencefoundation.org/keto-cta/ to download.

The observant reviewer will note a perhaps unusually low inter-metric coupling. What could this mean? Well, as I’ve said before and will say again, the science is never settled. This is an evolving story. Stay tuned.

Conclusion

As researchers, we understand that scientific progress is rarely linear or universally accepted on first pass. The spirited debate surrounding this paper reflects how deeply people care about getting this right—and that’s a good thing. We’ve made mistakes, acknowledged them, and worked diligently with the journal and broader community to bring more clarity to the conversation.

Our commitment to transparency, open data, and scientific discourse remains unwavering. We welcome ongoing scrutiny and dialogue, not as a burden, but as a privilege that comes with being part of an evolving field. Moving forward, our team is more motivated than ever to pursue meaningful science, improve communication, and honor the complexity of human biology with both humility and curiosity.

Post-script