World Heart Podcast

Play episode

Dr. Jiao Luo

Postdoctoral Researcher, University of Copenhagen · Deputy Editor, Global Heart

Dr. Raul Santos

Co-Host · Editor-in-Chief, Global Heart · Albert Einstein Hospital; Associate Professor, Heart Institute, University of São Paulos

n this episode, we unpack Mendelian Randomization (MR)—a powerful approach to causal inference that uses genetic variants as “natural experiments” to test whether a risk factor truly causes cardiovascular disease. Host Pablo Perel and co-host Dr. Raul Santos speak with Dr. Jiao Luo about when MR shines, where it fails, and how clinicians can interpret MR results alongside trials and biological plausibility. The conversation also clarifies why some popular biomarkers (e.g., CRP) may predict risk without being causal, and why raising HDL hasn’t translated into fewer events.

Download Episode

Follow our podcast

Apple Podcasts


Spotify


Deezer


Amazon Music


Podcast Index


Pocket Casts

Key Takeaways

Mendelian Randomization, in plain language

• MR uses inherited genetic variants (randomly allocated at conception) as instruments to test whether exposures (e.g., LDL-C, alcohol use tendencies) cause outcomes (e.g., ASCVD).

• Think of it as “lifelong randomization,” complementary to, not a replacement for, clinical trials.

The three core assumptions (must-haves)

1. Relevance: variants strongly predict the exposure (e.g., variants that robustly raise LDL-C).

2. Independence: variants aren’t associated with confounders (e.g., BMI).

3. Exclusion (no horizontal pleiotropy): variants affect the outcome only via the exposure of interest. Violating these breaks causal inference.

Common pitfalls & how to spot them

• Weak instruments → biased, imprecise estimates (look for instrument strength, e.g., F-statistics).

• Horizontal pleiotropy → variants act through other pathways; requires sensitivity analyses.

• Population mismatch → instruments discovered in European datasets don’t always transport to other ancestries.

• Linear-effect assumptions & timing → MR reflects lifelong average effects; drug effects over weeks/months may differ.

• Over- or under-interpretation → negative MR doesn’t automatically mean a drug can’t work; instruments might miss the right pathway.

Clinically meaningful examples

• HDL-C: Higher genetically predicted HDL-C is not causally protective—consistent with failed HDL-raising trials.

• CRP: Genetically higher CRP isn’t causal for ASCVD; CRP remains a useful risk marker, not a therapeutic target.

• LDL-C: Genetically higher LDL-C raises ASCVD risk—aligns with trial evidence for LDL-lowering.

How clinicians can quickly appraise an MR paper

• Check instrument strength and reported diagnostics.

• Look for multiple sensitivity analyses addressing pleiotropy.

• Sanity-check effect sizes versus trials/observational data.

• Ask: does the biological pathway make sense?

• Favor triangulation: MR + trials + mechanistic plausibility pointing in the same direction.

Global perspective: diversity & equity matter

Most genetic instruments come from European-ancestry cohorts. Building inclusive datasets and interpreting results in context are essential for globally relevant cardiovascular science and equitable care.

Resources & Further Reading

• Mendelian Randomization: Concepts & Applications – overview review (Deputy Editor Sanne/Sonny Peters).

• Unraveling the links between immune cells, lipids, and CVD – an MR study exploring inflammation–lipid interplay.

Stay Informed and Take Action

• Subscribe on your favorite platform above.

• Read the featured Global Heart pieces mention above