The keto/LDL story is a great example of the limitations of blood tests when studied in isolation, and what the future of medicine with LLMs will look like. Conventional teaching is to keep your LDL, or “bad” cholesterol, low. So when a pattern emerged of young, otherwise healthy people on keto diets experiencing massive spikes in their LDLs – to the levels of highest risk for heart attacks – it captured a lot of interest.

The people on keto diets who had the most dramatic increases in their “bad” cholesterol were actually the leanest and often quite fit. And the expected lab pattern accompanying high LDL in patients who develop coronary artery disease (high triglycerides and low HDL) wasn’t present. Their other cardiovascular and metabolic markers looked excellent. Were they really as high risk as their sky high LDL levels would suggest?

Several studies have explored this question. Most recently, the KETO-CTA trial, a single-cohort longitudinal study, examined the change in non-calcified coronary plaque volume over a year in 100 of the healthiest patients among the keto-diet-induced high LDL group. The results were sobering. Some of these patients had significant progression as feared, meaning that on the keto diet, even these healthy participants’ arteries were narrowing more quickly than expected. But others, also with high LDL on the keto diet, had little progression.

We don’t know why some people on keto diets are spared the negative atherogenic impacts expected from their extremely high LDL. A high LDL from a keto diet may be benign for someone at a sufficiently low risk for cardiovascular events. But if even the super metabolically healthy people chosen for the KETO-CTA trial did not all meet that standard, who is safe?

This story is a great example of tensions to come as more health optimizers start to poke holes in conventional medical advice. Many people will order blood tests on Function Health or Superpower - if they have a high LDL, here's how they could be led astray: It would be easy for an overly optimistic keto blog to quote this study as proof that the keto diet is harmless if you’re lean and fit. It would be easy for doctors to pat each other on the back from this study, that they were right all along that LDL accelerates plaques, which almost everyone has.

Both would be wrong. LDL is a surrogate marker. It doesn’t directly measure how long the arteries to your heart will stay happily unobstructed. Surrogates often work. But they have edge cases where they’re wrong, as this keto example shows quite convincingly.

The good news is, for the people hungry for the ground truth instead of polemics, more data and diverse individual stories are available than ever before. We don’t need to be nihilistic that with surrogate markers there is no ground truth. We just have to go deeper than the headlines. In this case, we got better answers through imaging, stratification, and by combining with tests of function.

Keto diets’ unique effect on lipids, aka diets are still not one-size-fits-all. Diet dangers, mitigated with right data.

The keto/LDL story is a great example of the limitations of blood tests when studied in isolation, and what the future of medicine with LLMs will look like. Conventional teaching is to keep your LDL, or “bad” cholesterol, low. So when a pattern emerged of young, otherwise healthy people on keto diets experiencing massive spikes in their LDLs – to the levels of highest risk for heart attacks – it captured a lot of interest. Doctors counseled these patients to go off their keto diets immediately. A diet that dramatically increases risk of heart attacks is obviously not going to get any doctor’s endorsement. But the truth turns out to be not so simple.

The people on keto diets who had the most dramatic increases in their “bad” cholesterol were actually the leanest and often quite fit. And the expected lab pattern accompanying high LDL in patients who develop coronary artery disease (high triglycerides and low HDL) wasn’t present. Their other cardiovascular and metabolic markers looked excellent. Were they really as high risk as their sky high LDL levels would suggest?

Several studies have explored this question. Most recently, the KETO-CTA trial, a single-cohort longitudinal study, examined the change in non-calcified coronary plaque volume over a year in 100 of the healthiest patients among the keto-diet-induced high LDL group. The results were sobering. Some of these patients had significant progression as feared, meaning that on the keto diet, even these healthy participants’ arteries were narrowing more quickly than expected. But others, also with high LDL on the keto diet, had little progression. Other case reports have observed rapid coronary artery disease progression in people with preexisting plaque disease. The keto diet with associated increases in LDL does seem dangerous if there is already plaque buildup.

The hypothesis becomes: is it safe if there isn’t? We don’t know why some people on keto diets are spared the negative atherogenic impacts expected from their extremely high LDL. A high LDL from a keto diet may be benign for someone at a sufficiently low risk for cardiovascular events. But if even the super metabolically healthy people chosen for the KETO-CTA trial did not all meet that standard, who is safe?

LDL is a surrogate marker. It doesn’t directly measure the health outcome: how long the arteries to your heart will stay happily unobstructed. Surrogates correspond with outcomes: generally. But they have edge cases where they’re wrong, as this keto example shows quite convincingly.

This story is also a great example of tensions to come as more health optimizers start to poke holes in conventional medical advice. Many people will order blood tests on Function Health or Superpower, and many of them will get lousy guidance on a high LDL. It would be easy for an overly optimistic keto blog to quote this study as proof that the keto diet is harmless if you’re lean and fit. It would be easy for doctors to pat each other on the back and say they were right all along that LDL accelerates plaques, which almost everyone has.

Both would be wrong. There is ammunition for both sides to say the other is stupid. Both are wrong and potentially dangerous if they entrench without nuance.

The good news is, for the people hungry for the ground truth instead of polemics, more data and diverse individual stories are available than ever before. We don’t need to be nihilistic that with surrogate markers there is no ground truth. We just have to go deeper than the headlines. In this case, we got better answers through imaging, functional testing, and by combining with other markers.