Patients attending stroke prevention clinics and a premature atherosclerosis clinic at University Hospital in London, ON, Canada were recruited to the study. They completed a dietary questionnaire, provided stool samples and had blood drawn to measure plasma levels of metabolites produced by the intestinal bacteria.

Patients were phenotyped by their residual score in linear multiple regression with measured carotid plaque burden as the dependent variable and coronary risk factors were predictors. The residual score essentially represents the distance off the regression line of predicted plaque. They were grouped into three categories: Unexplained atherosclerosis (with more plaque than predicted by risk factors; residual score >2); Explained (the amount of plaque predicted by risk factors, residual score >-2 and <2); and Protected (less plaque than predicted by risk factors, residual score <-2).

DNA was extracted from stool samples in the lab of Dr. Allen-Vercoe at University of Guelph. RNA makeup of the intestinal microbiome was assessed in the lab of Dr. Gregory Gloor at Western. Plasma levels of trimethylamine n-oxide, p-cresylsulfate, hippuric acid, p-cresyl glucuronide, pheny acetyl glutamine and phenyl sulfate were measured by ultra-performance liquid chromatography coupled to quadrupole time-of-flight mass spectrometry in the lab of Dr. Bradley Urquhart at Western.

Nutrient intake over the past year was calculated at the Harvard School of Public Health from the 131 item self-reported and semi-quantitative Harvard Food Frequency Questionnaire (FFQ).

Estimated glomerular filtration rate was calculated from the Chronic Kidney Disease Epidemiological (CKD-EPI) equations.

Inclusion Criteria:

• Patients attending stroke prevention clinics and a Premature Atherosclerosis Clinic at University Hospital, London, ON, Canada,
• with measurements of carotid plaque burden and the risk factors used in the linear regression model.
• Willing to consent to the protocol approved by the Ethics board

Exclusion Criteria:

• Missing data on variables used in the regression model,

male, female

as above

It will depend on requests.