Overview of #402 – NMR blood analysis with Jim Otvos, Ph.D.
Peter Attia interviews Jim Otvos, the biophysical chemist who pioneered NMR-based lipoprotein testing and commercialized LDL particle number measurement. The conversation traces how a mislabeled “cancer test” led to a breakthrough in blood analytics, then expands into how a single NMR blood sample can reveal much more than standard cholesterol panels: LDL particle burden, insulin resistance risk, chronic inflammation, and a composite mortality-risk signal called MVX. A major theme is that particle number and metabolic state often matter more than cholesterol mass alone.
Key Themes and Main Takeaways
1) NMR blood testing started as a cancer “false lead”
- Otvos’ team investigated a 1986 New England Journal paper claiming NMR could detect cancer from blood plasma.
- They quickly found the signal was not cancer-specific.
- The signal came from lipoproteins in plasma, not cancer biology.
- This serendipitous finding led to a new way to quantify VLDL, LDL, and HDL particles using NMR.
2) Standard lipid panels measure cholesterol content, not particle number
- Typical lipid testing uses chemistry assays to measure cholesterol in lipoproteins.
- LDL-C is often calculated indirectly using formulas such as Friedewald, rather than directly measured.
- NMR instead estimates particle concentration:
- LDL-P = number of LDL particles
- Reported in nmol/L
- Otvos emphasized that particles are containers for cholesterol, and particle count can better reflect risk than cholesterol content alone.
3) LDL particle number often outperforms LDL cholesterol for risk management
- People with the same LDL-C can have very different LDL particle burdens.
- When LDL-C and LDL-P are discordant, cardiovascular risk tends to track with LDL-P, not LDL-C.
- This matters especially for treatment:
- A patient may achieve “good” LDL-C while still having too many LDL particles.
- That can justify more aggressive LDL-lowering therapy.
4) “Large, fluffy LDL is harmless” is a misleading idea
- Small dense LDL is associated with higher risk, but Otvos argues that the apparent danger is largely because small LDL usually means more LDL particles.
- Once you adjust for LDL particle number, particle size itself adds little or no risk information.
- He pointed to familial hypercholesterolemia as an important counterexample: patients can have large LDL particles and still be at very high risk.
5) Discordance between LDL-C and LDL-P often reflects metabolic dysfunction
- Discordance becomes more common with features of metabolic syndrome:
- central obesity
- elevated blood pressure
- elevated fasting glucose
- low HDL-C
- elevated triglycerides
- This suggests that LDL-C/LDL-P mismatch is often a marker of underlying insulin resistance and metabolic health issues.
6) LPIR: an NMR-derived score for insulin resistance
- Otvos’ team created the LPIR (Lipoprotein Insulin Resistance) score.
- It combines six NMR-derived lipoprotein subclass measurements into a 0–100 score.
- Higher LPIR = more insulin resistance.
- LPIR was validated prospectively for predicting future type 2 diabetes and appears to outperform simpler markers like triglyceride/HDL ratio and fasting insulin.
- The key idea: insulin resistance is the causal process; elevated glucose is the downstream manifestation.
7) GlycA is a stable NMR marker of systemic inflammation
- GlycA is an NMR signal reflecting glycan structures on acute-phase proteins.
- It acts as a marker of chronic, low-grade systemic inflammation.
- Compared with CRP, GlycA is:
- less volatile
- more stable over time
- often more predictive in risk models
- It correlates with inflammatory cytokines such as IL-6 and responds to anti-inflammatory treatment.
8) MVX: a composite marker of metabolic vulnerability and mortality risk
- Otvos described MVX (Metabolic Vulnerability Index), a composite score built from:
- small HDL particle concentration
- GlycA
- citrate
- branched-chain amino acids: leucine, isoleucine, valine
- MVX is designed to capture vulnerability to poor outcomes, especially mortality.
- In multiple cohorts, MVX strongly predicts:
- all-cause mortality
- short-term mortality in high-risk clinical populations
- mortality risk even in apparently healthy adults
9) MVX may reflect “metabolic frailty,” not just disease presence
- A major insight from the discussion:
- MVX may not simply predict who develops disease.
- It may predict who is more likely to die from whatever illness or stressor occurs.
- This makes it more like a marker of resilience vs. frailty.
- Otvos emphasized that this is especially striking because MVX can be elevated in people who look clinically healthy.
10) The most surprising result: MVX in healthy 30-year-olds
- In a cohort of healthy adults aged roughly 25–30 at baseline, MVX distribution looked surprisingly similar to older cohorts.
- Some young adults had very high MVX scores despite no known disease.
- Over decades of follow-up, MVX still predicted mortality.
- This suggests MVX may capture something set early in life, not merely the result of aging or overt disease.
Practical Clinical Implications
What NMR adds beyond a standard lipid panel
- LDL particle number can reveal risk missed by LDL-C.
- LPIR can identify insulin resistance before glucose rises.
- GlycA can show chronic inflammatory burden.
- MVX may help identify people with hidden vulnerability to adverse outcomes.
How clinicians may use this information
- For atherosclerotic risk, look beyond LDL-C when:
- LDL-C and LDL-P are discordant
- metabolic syndrome is present
- triglycerides are elevated and HDL-C is low
- For diabetes prevention, LPIR may identify risk earlier than fasting glucose or HbA1c alone.
- For broader prognostication, MVX may eventually help identify patients at higher short-term mortality risk or poor resilience before overt decline.
Important Nuances and Limitations
1) Better prediction does not always mean better treatment decisions
- Otvos noted that adding LDL-P to broad cardiovascular risk models doesn’t always dramatically improve total prediction, likely because existing models already include correlated variables like HDL-C and triglycerides.
- The clearest use case is risk management, not just risk prediction.
2) Biomarkers are not perfect causal explanations
- A high MVX score likely reflects a real vulnerability state, but the exact biology is still being worked out.
- Some signals may be markers of downstream biology rather than direct causal drivers.
3) Commercial and reimbursement barriers have limited adoption
- Otvos argued that NMR testing has been underused not because it lacks value, but because:
- reimbursement systems reward existing assays
- clinicians and payers are slow to adopt new diagnostics
- LabCorp’s acquisition of Liposcience reduced the push for broader deployment
- He warned that current NMR analyzers will eventually age out, and the technology may need new investment to continue expanding.
Notable Insights
- “Size doesn’t matter” for LDL once particle number is accounted for.
- LDL-P and ApoB are closely aligned conceptually, though NMR can offer additional subclass information.
- Insulin resistance is the causal driver; glucose is the late marker.
- GlycA provides a stable signal of chronic inflammation that can be clinically useful in a way CRP sometimes isn’t.
- MVX may be a marker of resilience and metabolic frailty, not just disease burden.
Bottom Line
This episode makes a strong case for moving beyond traditional cholesterol testing. Otvos argues that NMR can extract clinically meaningful information from a single blood sample about:
- atherogenic particle burden
- insulin resistance
- systemic inflammation
- mortality/frailty risk
The biggest takeaway is conceptual: in many cases, what matters most is not the amount of cholesterol carried in blood, but the number and behavior of the particles carrying it—and the broader metabolic state they reflect.