I read Gonzalez et al. with great interest, especially alongside the newly published RISSCI-2 study in Atherosclerosis
doi.org/10.1016/j.atheroscle…
The observation of a median 53% LDL-C reduction with ezetimibe is certainly intriguing.
But we should be very careful about what this study actually demonstrates.
The 14 patients were retrospectively selected because they had already shown a marked LDL-C response to ezetimibe. They were not consecutive, there was no control group, and the prevalence of this response among unselected patients with ketogenic diet-associated hypercholesterolemia is unknown. Again the same problem!
Therefore, 53% is not an estimate of the expected effect of ezetimibe in ketogenic hypercholesterolemia. It is the response observed in a selected group of hyperresponders.
That distinction is fundamental.
The mechanistic interpretation is even less certain. The authors propose increased dependence on intestinal cholesterol absorption, yet sterol-balance markers did not show a uniform hyperabsorption phenotype. Negative testing for monogenic FH or sitosterolemia does not establish that the LDL elevation was caused by the ketogenic diet, particularly when pre-diet lipid values were frequently unavailable and formal clinical FH criteria were not systematically applied.
This becomes especially relevant in light of the new RISSCI-2 study in Atherosclerosis. In a prospective dietary intervention using stable-isotope tracing, differences in saturated-fat absorption did not explain interindividual LDL-C responsiveness, and changes in markers of cholesterol absorption and synthesis were remarkably similar between high and low responders.
RISSCI-2 does not directly refute Gonzalez et al. The populations and hypotheses differ.
But it provides an important reminder:
A striking pharmacological response does not prove the mechanism responsible for the underlying hypercholesterolemia.
Even more importantly, a retrospective case series of 14 selected responders cannot be used to ask why clinicians should prescribe statins or PCSK9-targeting therapies instead of a “$10 drug.”
That comparison confuses hypothesis-generating evidence with randomized cardiovascular-outcomes evidence.
Selecting patients because they had an exceptional response and subsequently presenting that exceptional response as evidence for treatment selection is not precision medicine; it is selection bias until prospectively validated.
Ezetimibe is an excellent, inexpensive, evidence-based drug. That does not require inflating what this study demonstrates.
The correct conclusion is actually the one contained in the paper itself: this is an interesting hypothesis-generating observation deserving prospective confirmation.
Precision medicine starts with phenotype. But rigorous science starts with knowing what the study design allows us to claim.