Uploaded March 2026 | Updated September 2026, 1 week ago
For the written breakdown, see this letter: staycuriousmetabolism.substack.com/p/we-want-to-retract-our-own-paper?r=40ekz2
In this conversation, I sit down with my friend and colleague Dave Feldman to explain why we’re seeking to retract our own study. For context: on April 7th, 2025 we published results from a trial of ~100 individuals on low-carb diets whose LDL cholesterol dramatically increased—so-called “lean mass hyper-responders.” We set out to answer a critical question: does this group experience accelerated plaque progression, and what predicts that risk?
Our original headline finding—that LDL cholesterol and ApoB did not predict plaque progression—still stands. However, the plaque progression data itself was based on an AI-guided analysis conducted by the company Cleerly. After publication, we discovered that their analysis was performed inappropriately (unblinded) with data suggesting errors or bias in the analysis. Unfortunately, the company declined to reanalyze or validate the work. Meanwhile, multiple independent, properly conducted analyses failed to replicate their findings.
At that point, it became clear: we could no longer stand behind the original plaque progression results. Importantly, the updated evidence suggests this population may actually be at lower risk than initially implied by the initial anomalous Cleerly findings. In this discussion, we walk through exactly what happened—the data, the errors, and the decision to retract—with full transparency. For those who’ve followed this story, I hope this provides clarity, closure, and a glimpse of where the science goes next.
Chapters:
0:31 — Why We Want the Paper Retracted
1:18 — How This Research Began
3:49 — The Study, the Patients, and the Core Finding
5:59 — Where the Results Started to Break Apart
12:29 — Why We Didn’t See the Full Data Before Publication
20:54 — The Raw Dataset Review
28:03 — The Independent Analyses That Changed Everything
41:04 — The Participant Reanalysis Plot Twist
49:43 — The Next Phase of the Research
For the written breakdown, see this letter: staycuriousmetabolism.substack.com/p/we-want-to-retract-our-own-paper?r=40ekz2
In this conversation, I sit down with my friend and colleague Dave Feldman to explain why we’re seeking to retract our own study. For context: on April 7th, 2025 we published results from a trial of ~100 individuals on low-carb diets whose LDL cholesterol dramatically increased—so-called “lean mass hyper-responders.” We set out to answer a critical question: does this group experience accelerated plaque progression, and what predicts that risk?
Our original headline finding—that LDL cholesterol and ApoB did not predict plaque progression—still stands. However, the plaque progression data itself was based on an AI-guided analysis conducted by the company Cleerly. After publication, we discovered that their analysis was performed inappropriately (unblinded) with data suggesting errors or bias in the analysis. Unfortunately, the company declined to reanalyze or validate the work. Meanwhile, multiple independent, properly conducted analyses failed to replicate their findings.
At that point, it became clear: we could no longer stand behind the original plaque progression results. Importantly, the updated evidence suggests this population may actually be at lower risk than initially implied by the initial anomalous Cleerly findings. In this discussion, we walk through exactly what happened—the data, the errors, and the decision to retract—with full transparency. For those who’ve followed this story, I hope this provides clarity, closure, and a glimpse of where the science goes next.
Chapters:
0:31 — Why We Want the Paper Retracted
1:18 — How This Research Began
3:49 — The Study, the Patients, and the Core Finding
5:59 — Where the Results Started to Break Apart
12:29 — Why We Didn’t See the Full Data Before Publication
20:54 — The Raw Dataset Review
28:03 — The Independent Analyses That Changed Everything
41:04 — The Participant Reanalysis Plot Twist
49:43 — The Next Phase of the Research
