Mean Common or Mean Maximum Carotid Intima-Media Thickness as Primary Outcome in Lipid-Modifying Intervention Studies

Aim: Carotid intima-media thickness (CIMT) measurements are used as a disease outcome in randomized controlled trials that assess the effects of lipid-modifying treatment. It is unclear whether common CIMT or mean maximum CIMT should be used as the primary outcome. We directly compared both measurements using aspects that are of great importance in deciding which is most favorable for use in clinical trials. Methods: A literature search was performed (PUBMED, up to March 31, 2008). Fifteen trials with lipid-modifying treatment were identified that had information on both outcome measures. Common CIMT and mean maximum CIMT were compared on reproducibility, strength of relation with LDL and HDL cholesterol and congruency of their results (harm/neutral/beneficial) with data from event trials. Results: Findings showed that the reported reproducibility was high for both measurements, although a direct comparison was not possible. The relationship between the achieved LDL-C and HDL-C levels with CIMT progression was modest and showed no difference in magnitude between CIMT measurements. CIMT progression rates differed across carotid segments with the highest progression rates observed in the bifurcation segment. Treatment effects differed across carotid segments without a clear preference pattern. Trials using mean maximum CIMT progression more often (12 out of 15 studies) paralleled the findings of event trials in contrast to the mean common CIMT (11 out of 15 studies), a difference not reaching statistical significance. Conclusions: Based on the literature, with equal results for reproducibility (assumed), lipid relationship and congruency with event findings, but with treatment effects that differ across carotid segments that can not be predicted, the mean maximum CIMT as the primary outcome may be preferred in trials on the impact of lipid-modifying interventions. One advantage is that information on mean common CIMT can generally be obtained easily in protocols assessing mean maximum CIMT, but not the other way around