Peter Attia MDCardiovascular disease & why we should change the way we assess risk | The Peter Attia Drive Podcast
CHAPTERS
- 0:02 – 10:35
Why the 10-year risk model fails to prevent premature disease
Allan Sniderman explains why basing statin prevention on 10-year risk makes preventing premature events nearly impossible. He describes how atherosclerosis begins in early life and why risk is low but event numbers are high under 60, and proposes selecting patients by causes.
- •Risk calculators are driven mainly by age and sex, so prevention effectively starts at 55 to 60
- •Almost half of infarcts and strokes occur before 60 because there are far more people under 60
- •Disease is already advanced by the time treatment starts under the 10-year approach
- •A causal benefit model uses non-HDL or ApoB to project 20- to 30-year risk
- •A single risk number like 4.1% says little about an individual's real risk
- 10:35 – 14:33
Pathology of early atherosclerosis and what statins do
Peter Attia recalls his pathology professor's remark about first heart attacks being fatal and how the pathology textbook showed atherosclerosis in young accident victims. Sniderman adds that the whole artery is diseased and statins mainly prevent complications by lowering ApoB particle number.
- •Pathologists see many first heart attacks because a large share are fatal
- •Autopsy images showed atherosclerosis in people in their 20s and 30s
- •The whole artery is diseased, not just one lesion, and normal architecture can't be restored
- •Statins lower ApoB particle number so fewer particles get trapped in the wall
- 14:33 – 20:57
A primer on cholesterol, lipoproteins and ApoB
Sniderman explains what cholesterol is and how chylomicrons, VLDL and LDL particles carry fats, each with an ApoB protein. He argues that particle number matters more than cholesterol content because cholesterol reaches the artery wall only inside ApoB particles.
- •Cholesterol and triglycerides must be carried in particles since they don't dissolve in water
- •Chylomicrons carry ApoB 48; VLDL and LDL carry ApoB 100
- •VLDL particles shrink and become LDL particles
- •Two people with the same LDL cholesterol can have different particle counts and risk
- •Cholesterol enters the artery wall only within an ApoB particle
- 20:57 – 30:14
History of the particle-versus-cholesterol evidence and discordance analysis
Attia and Sniderman trace how the discordance between cholesterol and particle number became clear, from Lees and Krauss to Sniderman's 1980 angiography study. Discordance analyses consistently favor ApoB over non-HDL cholesterol, and they discuss cost objections to ApoB testing.
- •Early work in the 1970s showed LDL particle number didn't track cholesterol or triglycerides exactly
- •Sniderman's 1980 study found a marked ApoB difference between clean and diseased coronaries
- •Discordance analyses all point to ApoB as the better marker
- •ApoB was standardized in 1994 while HDL, LDL and triglyceride measurements are not
- •The cost argument is undermined by prices of about $2 to $2.50
- 30:14 – 42:58
Limits of standard panels, calculated LDL and remnant type III
Sniderman explains that LDL cholesterol is almost always a calculation and that triglycerides reflect particles rather than being atherogenic themselves. They then cover remnant type III dyslipoproteinemia, which shows high cholesterol and triglycerides with low ApoB and can't be diagnosed without measuring ApoB.
- •LDL is usually calculated, with at least eight methods, and direct assays haven't been validated in diseased patients
- •Triglyceride-associated risk reflects higher particle numbers, not triglycerides themselves
- •Type III arises when VLDL particles aren't converted to LDL and become very cholesterol-rich
- •It is treated with statins and/or fibrates and a diagnostic app uses cholesterol, triglycerides and ApoB
- •Sniderman measures a lipid panel, Lp(a) once, and ApoB in everyone
- 42:58 – 53:48
Resistance to ApoB and Mendelian randomization
Sniderman is pessimistic about guidelines changing despite Mendelian randomization evidence. They explain how Mendelian randomization uses genes fixed at conception to get closer to causality, and that it supports ApoB and not HDL as causal.
- •Prospective cohorts like Framingham are limited by confounding over time
- •Mendelian randomization uses genetic variation fixed at conception to approach causality
- •Studies show ApoB captures the information in triglycerides, LDL-C and HDL-C
- •CETP inhibitor trials and MR both indicate HDL is not causal
- •MR can mislead when variables are linked, as with triglycerides disappearing after adjusting for ApoB
- •No phenotype is complete without an ApoB measurement
- 53:48 – 1:07:19
Hypertension, smoking and how ApoB changes with age
Attia asks how hypertension and smoking compare with ApoB. Sniderman says the pathophysiology of hypertension is poorly understood and offers his own hunch about the proximal aorta, notes that ApoB tends to stay high once high at 35, and objects to smoking prioritizing statin treatment.
- •Sniderman says the cause of hypertension isn't known and offers a personal guess about loss of elasticity in the proximal aorta
- •Comparing hypertension and ApoB as causes is not the right question, in his view
- •People high in ApoB at 35 have about a 90 to 95% chance of staying high
- •ApoB rises with menopause but more data are needed
- •Rising coronary disease despite statins may reflect obesity and diabetes and ApoB-LDL-C discordance
- •Smoking raises calculated risk and so moves people up the treatment priority list, which Sniderman objects to
- 1:07:19 – 1:11:39
Deciding who to treat early, and family history
Sniderman discusses the costs of delaying intervention and argues that gains from starting at 35 concentrate in people with high non-HDL, around 20% of the population. Attia asks about patients with normal ApoB but early disease in the family, and Sniderman says individual decisions are easier than writing group rules.
- •A paper in Circulation examined the cost of delaying treatment to ages 35, 45 and 55
- •Gains from starting early are small when non-HDL is low and larger when it is high
- •About 20% of the population is at evident high risk and should consider treatment
- •Family history would factor in, especially as ApoB rises toward the 50th to 60th percentile
- •Statins aren't cost-free and Sniderman says the link with diabetes isn't fully understood
- 1:11:39 – 1:23:28
Coronary artery calcium: uses and limits
Sniderman and Attia explain what a CAC score is and discuss when it is informative. Both see a positive score in a young person as alarming, while Sniderman would not use a negative score to override a high ApoB because the disease can still be advanced and developing.
- •CAC detects calcified plaque, a sign of advanced disease, and positive scores raise risk
- •In older people the test is often positive and adds little
- •In those under 60 a positive score can help someone on the fence decide
- •A zero score doesn't mean no disease is present or developing
- •Attia describes a two-by-two framework of age and zero versus non-zero score
- •The 30-year horizon makes a zero CAC add little in a 45-year-old with high ApoB
- 1:23:28 – 1:26:45
Risk, time horizon and the missing confidence intervals
Attia and Sniderman debate whether a 5% 10-year risk is meaningful, using a plane crash analogy, and why a 30% 30-year risk is easier to grasp. Sniderman then points out that risk estimates are published without confidence intervals.
- •Sniderman says a one-in-three risk is a number people can understand, unlike 7.8% in a decade
- •Attia argues 5% is larger than people realize and the treatment burden is low compared with not flying
- •Sniderman asks what the confidence interval is on a 5% estimate and says such intervals are never published
- •He says clinical care has become an industry that doesn't include error
- 1:26:45 – 1:44:02
Consensus, guidelines and the loss of scientific challenge
Sniderman argues that the consensus process in medicine has dampened contention of ideas, with unanimous guideline recommendations hiding weaknesses. They discuss the FOURIER trial and an FH study as examples of differing interpretations, and Sniderman says his own work challenging the 10-year model hasn't changed the guidelines.
- •Science depends on experiments that test hypotheses and on competing views
- •Evidence-based medicine and consensus processes can make conventional views last too long
- •Unanimous recommendations by about 100 anonymous experts are insulated from criticism
- •Attia and Sniderman discuss different readings of the FOURIER trial and a FH genetic-risk study Sniderman believes has a methodological error
- •One European guideline negative on ApoB cited a single paper
- 1:44:02 – 1:56:59
Communicating uncertainty and guidelines to clinicians
Attia and Sniderman discuss how guidelines are written and used by busy doctors. Sniderman argues for not compromising on truth, presenting a range of expert opinion and teaching doctors to handle doubt, while Attia uses an appendicitis example and raises COVID debate.
- •Doctors often read only one-page summaries of long guidelines
- •Sniderman rejects the idea that doctors and patients need information dumbed down
- •Attia uses appendicitis diagnosis as an example of calibrating around asymmetric errors
- •Sniderman says loud or highly credentialed voices can carry the day in committees
- •They discuss COVID as a case where dissent was suppressed, with Sniderman saying debate among experts is needed but not shrieking
- 1:56:59 – 2:05:58
Why debate has faded and the cost argument against ApoB
Sniderman says the weakness lies inside guideline rooms where people become invested in the guidelines and consensus is prized. He says the 2018 American guidelines cited few references on ApoB and mischaracterized cost, and expresses sadness that ApoB may not be adopted.
- •People are now identified with guideline membership rather than their own science
- •The 2018 American guidelines cited only four references on comparing ApoB with LDL-C and non-HDL-C
- •Guidelines present ApoB as a cost issue though a cost analysis found a $10 test added about 0.01% to care costs
- •Clinical chemistry groups say ApoB is more accurately measured, which guidelines don't mention
- •Sniderman voices sadness that ApoB won't be adopted
- 2:05:58 – 2:12:12
Optimism about change and the nonlinear advance of knowledge
Attia challenges Sniderman's certainty that ApoB won't be adopted, using the stonemason analogy and the fall of HDL-C as examples. Sniderman says he'll keep fighting, talks about the privilege of his work, and Attia praises his curiosity across fields.
- •Sniderman notes he'll turn 80 and acknowledges his language was self-pitying
- •Attia says knowledge advances nonlinearly and truth generally wins in science over time
- •HDL-C is cited as a recent reversal of consensus
- •Sniderman says he has a new observation about blood sugar and lipids that he thinks is startling
- •Attia credits Sniderman's breadth of curiosity for his insights
- 2:12:12 – 2:25:35
The 30-year causal model, compounding benefit and necessary-but-not-sufficient causes
Attia asks how Sniderman arrived at the 30-year causal model, and Sniderman explains it is based on available data, causal selection, and measurement precision. They discuss compounding benefits of early treatment, with retirement savings as an analogy, and ApoB as necessary but not sufficient for atherosclerosis.
- •30 years was the longest period with reliable data, and the 10-year model falls down between about 30 and 60
- •Grouping by ApoB or non-HDL measures cause with less error than risk equations that explain about 20% of variance
- •Patients can start, stop or revisit therapy and see ApoB progress
- •Small early risk reductions compound over decades, like retirement returns
- •Stopping disease from developing is perfect prevention, while treating established disease is only partial
- •ApoB is necessary but not sufficient, making it a good prevention target
