Christoffer Baustad Nilsen 🐈‍

There is a remarkable difference between finding evidence that challenges your assumptions and finding evidence that conveniently reassures you that you are the exception. This is an interesting paper by Sniderman and colleagues, examining nearly 294,000 UK Biobank participants and finding that higher HDL cholesterol modifies the association between ApoB and cardiovascular events. The interaction is real, statistically significant, and worthy of further investigation. But the leap from an observational association to suggesting that an individual with an LDL of 574 mg/dL and ApoB above 350 might be immune to atherosclerosis is quite another matter. In fact, the authors specifically caution that the observed interaction should not be taken as firm clinical evidence that ApoB ceases to be atherogenic or doesn't require treatment when HDL is high. That's a rather important sentence to leave out when you're discussing immunity. A statistical model showing attenuation of risk at high HDL is not a clinical trial. Nor does it establish that raising HDL is protective, or that someone with extraordinarily elevated ApoB can safely ignore decades of exposure. We've spent years discovering that increasing HDL cholesterol pharmacologically doesn't reliably reduce cardiovascular events. HDL is an interesting biomarker, but biomarkers and causal mechanisms aren't interchangeable. And then there is the problem of extrapolation. The ordinary distribution of ApoB in UK Biobank is not populated by large numbers of otherwise healthy individuals walking around with an LDL approaching 600. A modeled relationship at the extremes of HDL doesn't validate a personal experiment at the extremes of ApoB. Which brings me to my friend Simon Hill, @theproof, who recently published something I found considerably more rigorous and intellectually open. Simon had coronary CT angiograms suggesting a remarkable 48% reduction in plaque volume over 16.5 months. He could have celebrated, announced that his diet had reversed atherosclerosis, and made an excellent marketing campaign out of it. Instead, he examined the scans using three different analysis platforms. One suggested substantial regression, another progression, and the third essentially no change. Rather than promoting the most flattering result, he openly discussed the uncertainty. That's what scientific curiosity looks like when the evidence doesn't cooperate with your expectations. Simon also has something that makes this discussion more than academic. His father had a myocardial infarction at 41 and his grandfather at 49. Simon is now 40. He has lived with the consequences of cardiovascular disease in his family, and it's understandable that he wants to reduce his risk as much as reasonably possible. He has even started a PCSK9 inhibitor despite having an ApoB around 70–80 without medication. We can debate whether targeting an ApoB below 40 with Repatha offers sufficient additional clinical benefit in someone with his particular risk profile. That's a legitimate discussion. But I applaud his desire to live longer and his willingness to question his own most favorable findings. I understand the emotional component of this better than most. My own coronary calcium score is 90. I take rosuvastatin and ezetimibe, achieved an LDL around 60, and have seen it fall further since starting tirzepatide. I have no particular desire to measure every cubic millimeter of plaque in my coronary arteries. I know there is atherosclerosis, I know what the evidence supports, and I am treating the risk factors I can change. Simon wants more information and more aggressive prevention. I am comfortable treating what I already know. Neither approach requires pretending that uncertainty has disappeared. We physicians spend our lives talking about hazard ratios, plaque volumes, confidence intervals, and relative risk reductions. Yet one disrupted coronary plaque with a superimposed thrombus can turn all those abstractions into a very final event. That's why I care more about preventing myocardial infarctions than winning a competition over whose scan looks better. The remarkable thing about the comparison is that both Simon and Norwitz have favorable personal findings. Simon asks whether his are reliable. Norwitz highlights an observational study that might explain why his own findings are reassuring. There is a lesson there about confirmation bias, and it applies to all of us, physicians included. The Sniderman paper deserves attention. It may teach us something important about HDL biology and cardiovascular risk. But it does not establish immunity from ApoB, and it certainly doesn't overturn the causal evidence supporting LDL lowering. Science progresses when we investigate the observations that don't fit our models, not when we promote the observations that make us feel invulnerable. After all, the goal isn't to prove that your arteries are special. It's to live long enough that they never become the reason you aren't.