We aimed to identify food intake patterns that operate via haemostatic and inflammatory pathways on progression of atherosclerosis among 802 middle-aged adults with baseline and 5-year follow-up ultrasound measurements of common (CCA) and internal carotid artery (ICA) intimal medial thickness (IMT). Food intake was ascertained with an FFQ. We derived food patterns using reduced rank regression (RRR) with plasminogen activator inhibitor 1 and fibrinogen as response variables. We explored the impact of various food pattern simplification approaches. We identified a food pattern characterised by higher intakes of less healthful foods (low-fibre bread and cereal, red and processed meat, cottage cheese, tomato foods, regular soft drinks and sweetened beverages) and lower intakes of more healthful foods (wine, rice and pasta, meal replacements and poultry). The pattern was positively associated with mean CCA IMT at follow-up (P = 0·0032), a 1 sd increase corresponding to an increase of 13 μm higher CCA IMT at follow-up, adjusted for demographic and cardiovascular risk factors. With increasing pattern quartile (Q), the percentage change in CCA IMT increased significantly: Q1 0·8 %; Q2 3·2 %; Q3 8·6 %; Q4 7·9 % (P = 0·0045). No clear association with ICA IMT was observed. All simplification methods yielded similar results. The present results support the contention that a pro-inflammatory and pro-thrombotic dietary pattern increases the rate of coronary artery atherosclerosis progression, independent of traditional cardiovascular risk factors. RRR is a promising and robust tool for moving beyond the previous focus on nutrients or foods into research on the health effects of broader dietary patterns.
(Received June 22 2009)
(Revised October 27 2009)
(Accepted November 11 2009)
(Online publication January 22 2010)
Abbreviations: CCA, common carotid artery; FS, full food pattern score; ICA, internal carotid artery; IMT, intimal medial thickness; IRAS, Insulin Resistance Atherosclerosis Study; PAI-1, plasminogen activator inhibitor 1; RRR, reduced rank regression; SI, insulin sensitivity