Serendipity of post-hoc surrogate marker research.

The prerequisites under which ‘novel’ compounds have to show cardiovascular benefit on top of statin therapy have raised the bar for clinical endpoint studies. B-mode ultrasound measurement of the carotid intima media thickness (cIMT) is the best validated imaging modality and most widely used surrogate endpoint in intervention studies to evaluate treatment efficacy in the progression of atherosclerosis. In contrast to the overwhelming evidence of epidemiological trials linking increased cIMT to cardiovascular risk, the evidence for a direct relationship between decreasing cIMT progression and lowering cardiovascular risk is, however, less clear. Taylor and colleagues performed a post-hoc analysis in the Arbiter 6-HALTS trial in which they explored the relationship between treatment effects and cIMT change in 159 patients randomized to ezetimibe. Ezetimibe is a cholesterol absorption inhibitor (CAI) which lowers LDL cholesterol (LDL-c) both as monotherapy (LDL-c reduction 17.2–22.3%) and on top of statin therapy (further LDL-c reduction 5–27%). The Arbiter-6 HALTS study was discontinued prematurely after an interim analysis at 14 months showing that niacin reduced cIMT progression whereas ezetimibe showed no change in cIMT. In a subsequent publication, the authors confirmed a neutral effect of ezetimibe on mean cIMT change (–0.0016+0.0024 mm; P 1⁄4 0.88) with a hint of a positive relationship between the cumulative ezetimibe exposure (defined as ‘dose × adherence × duration’) and cIMT progression in 161 patients randomized to ezetimibe. To explore these findings further, an additional subgroup analysis was conducted reporting that, despite a mean LDL-c decrease from 84+23 to 66+ 20 mg/dL, there was a graded increase in cIMT progression observed across quartiles of increasing LDLc reduction (P , 0.001). Similarly, a trend was observed between cumulative ezetimibe exposure and cIMT progression (P 1⁄4 0.051). Although interesting, the observations reported in this small cohort are in contrast to the consistent inverse relationship between the LDL-c reduction and cIMT progression shown in a large number of randomized controlled trials comprising a much larger number of patients (Figure 1). More importantly, the Ezetimibe and Simvastatin in Hypercholesterolemia Enhances Atherosclerosis Regression (ENHANCE) researchers reported a positive correlation between LDL-c lowering and cIMT change in 642 patients with familial hypercholesterolaemia, showing that after 24 months a greater LDL-c reduction resulted in less progression of cIMT in the simvastatin–ezetimibe group. The latter is corroborated by data from the Stop Atherosclerosis in Native Diabetics Study (SANDS) study, which also reported cIMT reduction in 499 diabetic patients following 36 months of aggressive lipid-lowering therapy with no difference in cIMT response between ezetimibe vs. non-ezetimibe users. Collectively, the bulk of the data does not support an ezetimibe-induced progression of atherosclerosis as depicted by post-hoc analysis in the Arbiter 6-HALTS study. Besides the paradoxical cIMT progression in the ezetimibe-treated patients, the Arbiter 6-HALTS trial also reported a rapid and highly significant cIMT reduction (–0.0102+0.0026 mm; P , 0.001) in the niacin group. Based on these promising effects of niacin on cIMT, the authors concluded that the Arbiter 6-HALTS results merited strong consideration for the use of niacin in patients with low HDL. However, the AIM-HIGH trial, including 3414 patients, was prematurely discontinued because of a lack of incremental benefit of niacin over placebo with regard to cardiovascular events. As a result, we are presently facing a small surrogate cIMT study with results opposite to those of the outcome trial. What are the consequences of the Arbiter 6-HALTS findings for the choice of surrogate markers in studies on cardiovascular disease? When evaluating surrogate marker studies, one should be aware of potential weaknesses of the study, with a special emphasis on the power of the study, as well as the analytical procedures and choice of endpoints. Imaging studies can then help to guide decisions on phase III trials. In view of this goal, premature discontinuation of a surrogate marker study should be strongly discouraged. Multimodality imaging including magnetic resonance imaging and positron emission tomography–computed