PROJECT SUMMARY/ABSTRACT The proposed study will accelerate the discovery of blood-based DNA methylation (DNAm) biomarkers for cannabis use phenotypes (lifetime [ever vs. never] use and recency and frequency of use), by leveraging existing data on 9,878 individuals across eight cohorts. Cannabis is the most commonly used illicit drug in the United States, with 45% of Americans reporting lifetime use and 15% reporting past-year use in 2017. Both adverse (e.g., cannabis use disorder, cognitive impairment, bronchitis, psychosis) and beneficial (e.g., therapeutic benefits for certain clinical conditions) effects have been reported. To understand the full spectrum of associated health effects, there is an urgent need to develop tools that can accurately quantify patterns of cannabis use across a lifetime, yet currently available biomarkers, with limited windows of detection, lack these attributes. DNAm is an excellent candidate for biomarker development, as it has the potential to differentiate acute from chronic exposure and timing, duration, and frequency of exposure. As stressed by the National Institute on Drug Abuse director, Dr. Nora Volkow, and colleagues, there is an “urgent need for biomarkers that reflect chronic drug exposure ...”; yet, biomarker research that “take[s] advantage of epigenomics and epitranscriptomics is in its infancy”. We propose to assemble a collection of existing datasets across eight cohorts, enabling the largest epigenome-wide association study (EWAS) analyses of any cannabis use phenotype to date (N = 9,878). In Aim 1, we will identify general DNAm biomarkers of lifetime cannabis use (i.e., observed DNAm differences that can be driven by genetics and/or exposure). To achieve Aim 1, we will conduct an EWAS meta-analysis of lifetime cannabis use, from which we will use penalized regression to train and validate multi-CpG predictive models (i.e., DNAm biomarkers). In Aim 2, we will identify genetically- vs exposure-driven biomarkers of lifetime cannabis use, independently of Aim 1, by taking a multi-stage approach to tease apart the underlying mechanisms driving the DNAm differences. Each type of biomarker can be uniquely informative, with general biomarkers possibly providing the greatest overall predictive ability, genetically driven biomarkers as a refined phenotype for genetic studies, and exposure-driven biomarkers for evaluating the possible impact of behavior modification on related health effects. In Aim 3, we will develop general DNAm biomarkers of persistent (i.e., DNAm changes detected in both recent and former users), transient (i.e., DNAm changes detected in only recent users), and heaviness of cannabis use effects. These biomarkers can enable more specific future evaluations of cannabis-related outcomes (e.g., adverse effects related to persistent DNAm changes) and potential treatment applications (e.g., to help monitor adherence, as informed by a combination of transient and persistent DNAm changes).