Time-related trends in variability of cIMT changes in statin trials

This brief article provides complementary data supporting the results reported in “Changing Characteristics of Statin-related cIMT Trials from 1988 to 2006” [1]. That article described time-related trends in baseline factors and study characteristics that may have influenced the variability of carotid intima media thickness (cIMT) endpoints (mean of mean and maximum common carotid artery [CCA]/cIMT) in published statin trials. In this brief report, additional details for the studies included in the analysis, and further supporting data, including mean of the maximum CCA/cIMT changes and subgroup data (mean and maximum CCA/cIMT) are provided. For the analysis, study-level data was extracted from 17 statin cIMT trials conducted during 1988–2006, selected on the basis of having at least one statin monotherapy arm in the absence of mixed therapy, and baseline- and study-end values for mean mean and mean maximum CCA/cIMT endpoints. The baseline mean CCA/cIMT, maximum mean CCA/cIMT and LDL-C levels, and annualized cIMT changes were estimated for the overall studies, those conducted before/after 2000, and in risk-based subgroups. Interestingly, all 8 studies conducted before 2000 were significant for cIMT change in which patients did not receive prior LLT; whereas after 2000, the results were more variable and in 4 of 6 trials that did not show a significant cIMT change, patients had received prior treatment. Baseline mean maximum cIMT and LDL-C levels, and annualized changes in studies conducted before 2000 were higher than those conducted after 2000, similar to the results reported in the original article for the mean mean cIMT endpoint. These findings were consistent across study populations of patients with CHD risk versus those without, and in studies with greater LDL-C reductions and with thickened baseline cIMT at study entry for both mean and maximum cIMT changes. Taken together, these results are consistent with trends in recent years toward greater use of lipid-lowering therapy and control of LDL-C that may have impacted the variability in the results of cIMT studies.


a b s t r a c t
This brief article provides complementary data supporting the results reported in "Changing Characteristics of Statin-related cIMT Trials from 1988 to 2006" [1]. That article described time-related trends in baseline factors and study characteristics that may have influenced the variability of carotid intima media thickness (cIMT) endpoints (mean of mean and maximum common carotid artery [CCA]/cIMT) in published statin trials. In this brief report, additional details for the studies included in the analysis, and further supporting data, including mean of the maximum CCA/cIMT changes and subgroup data (mean and maximum CCA/cIMT) are provided. For the analysis, study-level data was extracted from 17 statin cIMT trials conducted during 1988-2006, selected on the basis of having at least one statin monotherapy arm in the absence of mixed therapy, and baseline-and study-end values for mean mean and mean maximum CCA/cIMT endpoints. The baseline mean CCA/cIMT, maximum mean CCA/cIMT and LDL-C levels, and annualized cIMT changes were estimated for the overall studies, those conducted before/after 2000, and in risk-based subgroups. Interestingly, all 8 studies conducted before 2000 were significant for cIMT change in which patients did not receive prior LLT; whereas after 2000, the results were more variable and in 4 of 6 trials that did not show a significant cIMT change, patients had received prior treatment. Baseline mean maximum cIMT and LDL-C levels, and annualized changes in studies conducted before 2000 were higher than those Contents lists available at ScienceDirect journal homepage: www.elsevier.com/locate/dib conducted after 2000, similar to the results reported in the original article for the mean mean cIMT endpoint. These findings were consistent across study populations of patients with CHD risk versus those without, and in studies with greater LDL-C reductions and with thickened baseline cIMT at study entry for both mean and maximum cIMT changes. Taken together, these results are consistent with trends in recent years toward greater use of lipidlowering therapy and control of LDL-C that may

Value of the data
The relationship between baseline and study characteristics and the variability of cIMT change in statin cIMT trials over time has not been well-studied.
These data show time-related trends in baseline and study characteristics that may influence the variability of mean mean and mean max cIMT results in statin clinical trials.
Statin cIMT trials conducted before 2000 had greater cIMT baseline values, LDL-C reduction and annualized cIMT changes than studies after 2000.
These trends are consistent with increased statin treatment and management of LDL-C lowering over recent years, bringing into question the utility of cIMT as a surrogate marker in clinical trials given the high standard of background statin therapy.
These results highlight the need to consider such factors in the design of future cIMT trials.

Data
The predictive power of cIMT as a surrogate marker for the assessment of CVD risk reduction in lipid-lowering trials may be limited in some settings due to differences in study design, cIMT methods, and patient characteristics [2][3][4]. With increased use of lipid-lowering therapy over recent years, it is possible that patient characteristics and study designs of cIMT trials have also changed in a manner which could influence the utility of cIMT in assessing LLT treatment effects. This analysis was undertaken to better understand how baseline factors and study characteristics may influence the variability of cIMT changes in intervention trials.
A summary of the cIMT trials included in the analyses reported in the original article [1] and in this brief report are summarized in Table 1. All 8 of the studies that were conducted before year 2000 were significantly positive for cIMT change, and the patients in those trials did not receive prior lipidlowering therapy. However, the 9 studies conducted after 2000 were more variable for cIMT changes nn Study was statistically significant (po 0.05) for primary endpoint. n Study was non-significant (p 40.05) for primary endpoint. † MI or hospitalization for unstable angina 3 mon-5 yr before enrollment and a total serum cholesterol 4-7 mmol/L (155-271 mg/dL) after 2 mon on low-fat diet; 88% men, 75% infarction, 5% diabetics. ‡ LDL-C 120-190 mg/dL (3.10-4.92 mmol/l) w/0-1 CHD risk factors or 120-160 mg/dL (3.10-4.14 mmol/l) w/Z2 CHD risk factors and 10 yr risk o10%, HDL-C r60 mg/dL (1.55 mmol/l); max  with 6 of these not showing significant cIMT change, the majority of which (4 of 6 trials) included patients that had received prior lipid-lowering therapy. The baseline mean mean and mean maximum cIMT and LDL-C levels were also generally greater for those studies conducted before 2000 than after 2000. The durations of the trials were also longer (Z 2 years) in those conducted before 2000. Table 2 summarizes the mean maximum baseline CCA/cIMT level and annualized changes for the overall studies (1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006) and those conducted before and after 2000, as well as for various study populations. For the overall studies combined (1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006), the baseline mean maximum CCA/cIMT level was 1.0866 mm and the annualized rate of change for the mean maximum CCA/cIMT from baseline was À 0.0023 mm/yr (95% CI: À 0.0050, À 0004), indicating a statistically significant annualized change in cIMT. Baseline mean maximum cIMT levels were higher in populations of studies conducted before 2000 (1.2550 mm) than after 2000 (1.0181 mm), as well as in study populations of patients with CHD risk versus those without, and in those studies with greater LDL-C reductions, and those with thickened baseline cIMT. Annualized rates of change in maximum CCA/ cIMT levels were also larger for the combined study populations conducted before year 2000 ( À 0.0134 mm/yr) than after 2000 ( À0.0083 mm/yr), and for those studies in which patients had thickened baseline cIMT at study entry and also those which showed greater LDL-C reductions.
The baseline mean CCA/cIMT and annualized changes observed in pooled studies conducted before and after 2000 for various prespecified subgroups of interest are displayed in Table 3. Consistent with the findings in the overall analysis cohort, baseline cIMT levels and annualized change for mean CCA/ cIMT were generally greater in all subgroups assessed for the combined studies conducted before 2000 than after 2000. Similarly, mean max CCA/cIMT and annualized change in studies were also generally greater in all subgroups assessed in studies conducted before 2000 than after 2000 (Table 4).

Study design
This was an exploratory analysis of study-level data from statin treatment arms of published cIMT imaging trials, as previously described in the original study article [1]. Fig. 1 displays the method of trial selection for the analysis. Following a detailed review of cIMT trials in the literature, 24 statin trials were identified that had both baseline-and study-end measurements for mean of the mean CCA/cIMT (mean of all mean measurements on CCA or a single mean cIMT value when not available), and/or mean of the maximum CCA/cIMT (mean of all maximum mean measurements on CCA, or single maximum mean or bulb values when not available). Of these, 7 trials were excluded due to insufficient data or mixed lipid-lowering therapy in the statin arm, and 17 studies conducted during 1988-2006 were selected on the basis of having at least one statin monotherapy arm in the absence of mixed therapy, baseline-and study-end values for mean of the mean and mean of the maximum CCA/cIMT. The selected studies included a mix of placebo (ACAPS, ARBITER II, BCAPS, CAIUS, CAPTI-VATE, CASHMERE, KAPS, LIPID, METEOR, PLAC II, REGRESS) and active-controlled (ARBITER I, ASAP, RADIANCE I, RADIANCE II, ENHANCE, VYCTOR) studies (Table 1 and Fig. 1) [5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21]. There were 13 trials with mean mean CCA/cIMT as the primary endpoint and 12 with mean maximum CCA/cIMT endpoints; these were assessed in 2 separate, but similar analyses.

Data extraction and analysis
Only aggregate data in the published literature were collected from the studies. Demographic information (e.g., age and gender), study characteristics (study dates, CHD risk, heterozygous familial hypercholesterolemia [FH], baseline LDL-C, HDL-C and TG levels), and baseline-and study-end measurements for mean mean CCA/cIMT and mean max CCA/cIMT, and variability estimates were extracted from the publications. Baseline means of the mean CCA/cIMT and maximum CCA/cIMT levels were summarized for the overall population of studies, those conducted before and after 2000 (the median of enrollment dates for all studies), and studies that enrolled patients based on CHD risk,    Study-end values were reported as a change from baseline in primary variables (mean mean or mean max CCA/cIMT) indicating regression and/or progression of cIMT. Annualized rates of change (regression and/or progression) were extracted as treatment differences for placebo-controlled studies and as change from baseline in each individual treatment arm for active-control studies. For active-controlled studies with 2 or more monotherapy arms, each treatment arm was included in the analysis as a separate data point. For trials in which annualized rates of change was not provided, it was calculated by dividing the reported change in cIMT by the duration of the study in years.
Estimates of variance for the baseline and study-end values for mean mean and mean max cIMT were extracted from the publications. In most cases the standard deviation was reported for baseline values and the standard error (SE) was reported for study-end values. In some instances the SE was not reported, and it was calculated from the reported 95% CIs or p-values for the treatment difference (placebo-controlled) or change from baseline (monotherapy arms) in cIMT. The rate of cIMT change was standardized (annualized) for this analysis; however, the SE reported in the publication was used regardless of study duration. An overall pooled estimate for the baseline CCA/ cIMT value and annualized rate of cIMT change was calculated by weighting each individual study by the inverse of the square of its standard error which was extracted from the publications. This resulted in studies with small variability having more weight in the pooled values than studies with large variability. Overall pooled estimates for the baseline CCA/cIMT value and annualized rate of change for mean of mean CCA and mean of max cIMT, were calculated using a fixed-effects model, by weighting each individual study by the inverse of the square of its standard error; thus, studies with small variability had more weight in the pooled values than studies with large variability.

Source of funding
Merck Sharp & Dohme Corp., a subsidiary of Merck & Co., Inc.