On Saturday, March 14, 2026, a documentary called The Cholesterol Code premiered at CoSci in Las Vegas.¹ The notice retracting the headline paper from the study at its center is dated three days earlier, March 11, and appeared in the May issue of JACC: Advances.^2,3

The authors had publicly disclosed their retraction request before the premiere. On March 5, Nick Norwitz, a co-first author, wrote that they had "formally and repeatedly pressed" for it.⁴ They had already posted a blinded reanalysis of the same scans and believed it supported their central finding: LDL levels still didn't predict whose plaque grew.^4,5 The premiere went ahead under the tagline "The Science Is Not Settled."⁶

The distributor's promotional copy casts Dave Feldman, the man behind the study, as David against Big Pharma, corporate healthcare, and mainstream science, and asks whether we've been "fighting the wrong enemy all along."⁷ To me, the film's marketing captures what went wrong in this corner of the keto movement: a compelling story became more confident than the evidence behind it.

The research question was worth asking. But whether ApoB, the protein on every LDL particle and its plaque-building relatives, predicted one year of plaque change within this selected group is a different question from whether maintaining very high ApoB is safe for decades. Much of the controversy comes from treating those questions as interchangeable.

Disclosure: I'm chief medical officer of Vineyard, a telehealth obesity practice, and I consult for Viking Therapeutics and Novo Nordisk. I've criticized the lean mass hyper-responder hypothesis in print before, once leaning on a number that has since been retracted. More on that below.

Feldman's own LDL soared after he went low-carb.^8,9 In 2017 he named the pattern he kept seeing in lean people on ketogenic diets, very high LDL with high HDL and low triglycerides: lean mass hyper-responders.¹⁰ The question underneath it is fair: does LDL that doubles or triples on a diet in a lean, insulin-sensitive person carry the usual risk? Feldman founded the Citizen Science Foundation to pay for a study to find out, coauthored its papers, and co-wrote the film.^9–13

Studying lean mass hyper-responders was worthwhile. My criticism is of how the results were presented and how much reassurance people were asked to take from them. Everything below is on the record.