I'm often asked how human genetics can be used to validate drug targets. In our new @NatureCVR paper, we provide an end-to-end framework for IL-6 inhibition🧵 👉build a genetic proxy 👉demonstrate potential for cardiovascular benefit 👉assess expected & unexpected safety signals

🔗link to the paper @ScienceMagazine: https://t.co/uIruKApmPZ 📊online resource with disease associations: https://t.co/MUZeOrSGVP

2⃣The cumulative effect of age, sex, BMI, and disease status in the variance of each protein across the proteome For most proteins, the explained variance for each protein is small — <50%.

Two additional interesting aspects of this analysis 1⃣The profiling of changes of the blood proteome from childhood to adulthood across 4 visits in 100 subjests (4, 8, 16, 24 years) I'm not sure I've seen this anywhere before.

Most blood proteomics analyses of human disease focus on a single disease endpoint missing assessment of specificity. This analysis of 8,262 individuals across 59 diseases offers an atlas of Olink proteomic signatures across the disease spectrum.

The fact that the results are driven by "unspecified dementia", which is mostly vascular + Alzheimer's, but there are almost no effects on either vascular or Alzheimer's dementia seems a bit off to me. Plus some of the effect estimates seem implausible: e.g. ORs of 0.06

🔗paper: https://t.co/SFtdnowcfx 📊web portal: https://t.co/1ozAR7gwrd The score was trained on data from Open Target: https://t.co/dAL3G1twXQ

Human genetic evidence can inform drug development at several levels. A new scoring framework used genetic evidence to rank probability of side-effects for target-outcome pairs. The results for 15,139 protein-coding genes and 499 drug side effects are summed in a web portal.

🔗link to paper: https://t.co/sR1L9sQq4o

CheMBL is a comprehensive open-access database of approved drugs and drugs in clinical development. Whenever I tried it, the data structure always felt complex. This paper sums processes developed over 15 years to curate and integrate CheMBL data👇

5 advances in medicine this week 1. One-and-done cholesterol therapy This is an epigenetic editor of PCSK9 — to silence its expression with a one-off dose and lower cholesterol (here in macaques) This complements existing phase 1 human data from Verve Therapeutics for a

...not a new concept though. This is for example the approach we used to proxy IL6 inhibition in our recent paper:

🔗link to the paper: https://t.co/UyZiSL6EeL

I agree that using downstream readouts (even if not causal for the outcome of interest) is the most secure way of obtaining reliable instrument. But positive/negative controls are always necessary. Too many traps when using gene expresion or protein abundance as proxy.

Nice read on cis-Mendelian randomization and the use of downstream readouts as proxy exposures for perturbations of a gene (or its encoding protein) of interest.

...and mechanisms of action

Anti-inflammatory drugs for atherosclerotic cardiovascular disease in clinical development

🔗link to the paper 🔓: https://t.co/QZJAeNZK9R

Stroke prevention is more nuanced than that of atherosclerosis manifestations in other vascular beds due to heterogeneity in underlying etiology. In this paper we discuss the implications of recent trials for stroke prevention, as well as future perspectives for under

If you're up for a niche Sunday read on challenges in translating anti-inflammatory therapies for atherosclerotic stroke prevention —in line with genetic data, have a look at our review in Neurology (@GreenJournal) with @ArKatsanos Philip Melton @LukaZett and @anna_kopczak

🔗link to study: