Because saliva 🧪 is more common than blood 🩸 in large pediatric 👶👧 cohorts, we advocate for more saliva-customized algorithms!

So, we are losing some between-person 👥signal when applying blood algorithms to saliva. How much depends on the algorithm. While this moderate level of similarity may be useful for some research applications, it probably lacks the accuracy needed for clinical or commercial use💡

We found saliva-blood cross-tissue ICCs were mostly moderate for 2nd/3rd gen clocks and DNAm of physiology and Epigenetic-g. PCGrimAge (r=0.76) was highest, then PCPhenoAge (r=0.72) and Epigenetic-g (r=0.69). 1st gen clock acceleration showed poor similarity.

To see if we get a similar ranking of people across these tissues, we calculated the cross-tissue ICCs and compared them with the within-tissue ICCs from technical replicates in other studies ⬇️

We pooled data from 3 studies with saliva 🧪 and blood 🩸 DNAm (n=107). In each tissue, we computed 3 generations of epigenetic clocks and DNAm measures of CRP, BMI, and cognition (epigenetic-g), adjusting for cell count variation, as we do in our saliva studies

Most kids prefer to spit in a tube over a needle! But these algorithms were made in blood🩸so are we measuring something similar in saliva? Our twin👯and longitudinal work is promising, but we’re unsure how much signal we’re losing ➡️ jamanetwork.com/journals/jam...

In adults, epigenetic clocks become key biomarkers of age-related health. Second- and third-generation clocks, plus DNAm measures of CRP, BMI, and cognition, are sensitive to social determinants of health in children ➡️ journals.sagepub.com/doi/full/10.... www.nature.com/articles/s41...