The aim of this study was to evaluate the effect of atorvastatin on lipid lowering, cardiovascular (CV) events, and adverse events in women compared with men in 6 clinical trials. In the Incremental Decrease in End Points Through Aggressive Lipid Lowering (IDEAL) trial (atorvastatin 80 mg vs simvastatin 20 to 40 mg), the Treating to New Targets (TNT) trial (atorvastatin 80 vs 10 mg), the Stroke Prevention by Aggressive Reduction in Cholesterol Levels (SPARCL) trial (atorvastatin 80 mg vs placebo), and the Collaborative Atorvastatin Diabetes Study (CARDS), the Anglo-Scandinavian Cardiac Outcomes Trial (ASCOT), and the Atorvastatin Study for Prevention of Coronary Heart Disease Endpoints in Non-Insulin-Dependent Diabetes Mellitus (ASPEN) (atorvastatin 10 mg vs placebo), lipid changes on treatment were compared between genders with studies grouped by dose. The association of on-study low-density lipoprotein (LDL) cholesterol and CV events by gender was evaluated in the combined studies and the impact of gender on adverse events in each study separately. Major CV events occurred in 3,083 of 30,000 men (10.3%) and 823 of 9,173 women (9.0%). Changes in lipids were similar in women and men. Major CV events were associated with gender-specific quintiles of on-treatment LDL cholesterol for women and men. In women, LDL cholesterol was a significant predictor of stroke, but not in men. Discontinuation rates due to adverse events were higher in women in 4 of 6 trials, but in only 1 trial was a significant treatment-gender interaction seen. Myalgia rates were slightly higher in women in both statin and placebo groups. In conclusion, the response of women to atorvastatin was similar to that of men, with slightly more discontinuations due to adverse events. Higher on-treatment LDL cholesterol was significantly associated with more CV events in both genders, but the association was stronger for stroke in women and for coronary heart disease death in men.
Lipid-lowering therapy with statins decreases cardiovascular (CV) events and mortality in a variety of patient populations and clinical scenarios. Fewer women than men have been enrolled in statin trials, and whether statins provide benefit to certain subsets of women has been controversial. For example, a meta-analysis of 6 trials including 11,435 women without CV disease published a decade ago showed no benefit for statins for any of the CV end points, although benefit was seen for secondary prevention. In a meta-analysis of statins for primary prevention in women published 6 years later with larger numbers of subjects, the relative risk for CV events for statin-treated women was 0.63 (95% confidence interval [CI] 0.49 to 0.82, p <0.001), with a trend toward a reduction in total mortality (relative risk 0.78, 95% CI 0.53 to 1.15). Two recent meta-analyses came to opposite conclusions. Gutierrez et al concluded that statins for secondary prevention reduced total mortality and stroke in men but not women, while Kostis et al showed similar benefits for men and women in primary and secondary prevention, including similar reductions in total mortality.
In most statin trials, adverse events have not been reported according to gender and have not included data on each individual patient. Some evidence suggests that statin discontinuation rates are higher in women, and despite objective evidence, women are generally considered to be more likely than men to have side effects related to statins. Given these considerations, the purpose of this study was to evaluate the impact of female gender on lipid lowering, CV events, and adverse events (AE) in 6 large randomized clinical trials using patient-level data. These trials included atorvastatin at high and low doses in the settings of primary and secondary prevention: Treating to New Targets (TNT), Incremental Decrease in End Points Through Aggressive Lipid Lowering (IDEAL), Stroke Prevention by Aggressive Reduction in Cholesterol Levels (SPARCL), the Collaborative Atorvastatin Diabetes Study (CARDS), Atorvastatin Study for Prevention of Coronary Heart Disease Endpoints in Non-Insulin-Dependent Diabetes Mellitus (ASPEN), and the Anglo-Scandinavian Cardiac Outcomes Trial (ASCOT).
Methods
The main results of the 6 trials included in this analysis have been published previously and are summarized in Table 1 . All trials were randomized, double-blinded, and had placebo or active treatment comparators. Follow-up was open label, with blinded end point evaluation in IDEAL and blinded in the other 5 trials. The primary end points differed among the 6 trials, and for this study, major CV events were defined as CV death, myocardial infarction, resuscitated cardiac arrest, and stroke. The analysis plan for this study was designed prospectively to answer questions related to gender differences with statins in low-density lipoprotein (LDL) cholesterol lowering, CV events, and discontinuation rates due to AEs and creatine kinase (CK) elevations.
| Variable | TNT | IDEAL | SPARCL | CARDS | ASPEN | ASCOT-LLA |
|---|---|---|---|---|---|---|
| Study Drug | A 80 mg | A 80 mg | A 80 mg | A 10 mg | A 10 mg | A 10 mg |
| Comparator | A 10 mg | S 20-40 mg | Placebo | Placebo | Placebo | Placebo |
| Patients | 10,001 | 8,888 | 4,731 | 2,838 | 2,410 | 10,305 |
| Women | 1,902 (19.0%) | 1,701 (19.1%) | 1,908 (40.3%) | 909 (32.3%) | 811 (33.7%) | 1,942 (18.8%) |
| Mean FU | 4.9 years | 4.8 years | 4.9 years | 3.9 years ∗ | 4 years | 3.3 years ∗ |
| Entry criteria | CHD, LDL 130-250 mg/dl | History of MI | Stroke or TIA 1-6 months | Diabetes + another RF, no CAD | Diabetes | Hypertension + 3 RFs, no CAD |
| Primary endpoint | CHD death, MI, stroke, cardiac arrest | CHD death, MI, cardiac arrest | Fatal or non-fatal stroke | CHD death, MI, UA, stroke cardiac arrest, PCI, CABG | CHD death, MI, UA, stroke cardiac arrest, PCI, CABG | CHD death, MI |
| Baseline LDL-C (mg/dl) | 98 | 121 | 133 | 117 | 113 | 133 |
| On-treatment LDL-C (mg/dl) | 77 vs 101 | 81 vs 104 | 73 vs 129 | 75 vs 119 | 79 vs 113 | 87 vs 131 |
| Event rates | 8.7 vs 10.9% | 9.3 vs 10.4% | 11.2 vs 13.1% | 5.8 vs 9.0% | 13.7 vs 15.0% | 1.9 vs 3.0% |
| HR (95% CI) | 0.78 (0.69-0.89) | 0.89 (0.78-1.01) | 0.84 (0.71-0.99) | 0.63 (0.48-0.83) | 0.90 (0.73-1.12) | 0.64 (0.50-0.83) |
To assess the impact of gender on lipid lowering, changes from baseline in LDL-C, high-density lipoprotein (HDL) cholesterol, triglycerides, and non-HDL cholesterol over time by treatment were analyzed. Within each gender, treatment comparisons were performed using an analysis-of-covariance model containing baseline and treatment. Treatment-by-gender interactions were computed to assess consistency of treatment effect by gender. IDEAL and TNT were pooled and presented by treatment (atorvastatin 80 mg, simvastatin 20 to 40 mg/atorvastatin 10 mg). SPARCL was summarized separately, and CARDS, ASCOT, and ASPEN were pooled (atorvastatin 10 mg vs placebo for all 3 trials). To assess the association of LDL and HDL cholesterol on CV events by gender, the 6 studies were pooled. We evaluated major CV events, coronary heart disease (CHD) death, nonfatal myocardial infarction, stroke, and CV mortality. Analysis was performed by month 3 LDL cholesterol gender-specific quintiles for all studies except ASCOT, in which month 6 LDL cholesterol was used because month 3 lipids were not collected. Subjects with events occurring before month 3 LDL cholesterol or month 6 LDL cholesterol in ASCOT were excluded from the analysis. Cox proportional-hazards models adjusting for study and treatment were used and hazard ratios (HRs) were computed comparing each quintile with the lowest LDL cholesterol quintile. The p value for trend across the 5 quintiles was computed. The analysis was performed for men and women separately. To examine the consistency of LDL cholesterol effect on events by gender, the interaction between gender and LDL cholesterol quintiles was computed. This analysis was repeated using month 3 HDL cholesterol quintiles.
The impact of gender on discontinuation rates was examined separately for each of the 6 studies. A summary of discontinuation by treatment, reasons for discontinuations, and the most frequent AEs associated with discontinuation was generated. Statistical comparisons were performed for men and women using logistic regression comparing the difference in rates between treatments, and the treatment-by-gender interaction was computed for overall discontinuations due to AEs. Predictors of discontinuation were assessed for men and women separately.
An analysis to assess the effect of gender on CK elevations and AEs was performed for the 5 trials for which this information was available (CK levels were not measured in ASCOT). CK levels before treatment were categorized, and shifts in CK elevations were summarized at the maximum CK level during the study. Rates of discontinuations due to AEs and treatment emergent myalgia were summarized for the baseline CK subgroups.
Results
Selected features of the 6 trials are listed in Table 1 . Of the 39,173 patients enrolled in the 6 trials, 9,173 (23.4%) were women. As listed in Table 2 , women were older (63.3 vs 61.6 years), were more likely to have hypertension and diabetes, weighed less, and had higher LDL and HDL cholesterol levels at baseline. Mean follow-up for 4 of the trials was 4-5 years but was shorter in CARDS and ASCOT–Lipid Lowering Arm (ASCOT-LLA) because they were stopped early for benefit. Major CV events occurred in 3,083 of 30,000 men (10.3%) and 823 of 9,173 women (9.0%).
| Variable | Women (n=9,173) | Men (n=30,000) | P value |
|---|---|---|---|
| Age (years) | 63.3±9.4 | 61.6±9.1 | <0.0001 |
| Diabetes mellitus | 3,269 (35.6%) | 7,870 (26.2%) | <0.0001 |
| Hypertension | 6,467 (70.5%) | 18,686 (62.3%) | <0.0001 |
| Current smoker | 1,939 (21.1%) | 6,450 (21.5%) | 0.75 |
| Body mass index (Kg/m 2 ) | 28.6±5.2 | 28.1±4.1 | <0.0001 |
| Body weight (Kg) | 74.6±14.5 | 86.1±14.0 | <0.0001 |
| Systolic BP (mmHg) | 143.7±22.4 | 142.8±22.5 | 0.0015 |
| Diastolic BP (mmHg) | 82.1±11.6 | 84.4±12.2 | <0.0001 |
| Total cholesterol (mg/dl) | 207.5±34.5 | 194.4±34.2 | <0.0001 |
| LDL-C (mg/dl) | 122.0±31.1 | 117.7±30.1 | <0.0001 |
| HDL-C (mg/dl) | 55.0±14.5 | 46.8±11.8 | <0.0001 |
| Triglycerides (mg/dl) | 153.3±78.3 | 150.2±82.3 | 0.0017 |
| ApoA1 (mg/dl) | 124.0±73.1 | 95.0±69.0 | <0.0001 |
| ApoB (mg/dl) | 93.8±55.8 | 78.3±57.7 | <0.0001 |
Changes in lipids were generally similar in women and in men, as shown in Figure 1 for LDL cholesterol. High-dose atorvastatin lowered LDL and non-HDL cholesterol compared with the moderate dose slightly more in men than in women in the pooled TNT and IDEAL patients (p = 0.003 for treatment-by-gender interaction at 3 months for LDL cholesterol), but the absolute difference was small. No gender difference was seen with atorvastatin 80 mg in SPARCL. In contrast, atorvastatin 10 mg lowered LDL and non-HDL cholesterol compared with placebo slightly more in women than in men in the pooled CARDS, ASPEN, and ASCOT-LLA patients (p = 0.02 for treatment-by-gender interaction at 6 months for LDL cholesterol), but this difference was also small and did not persist consistently at subsequent visits.
To consider the impact of discontinuation rates on lipid changes, the analysis was repeated after excluding all evaluations that occurred after the last treatment date. Overall, the decreases in LDL were slightly larger, and for IDEAL and TNT, the treatment-by-gender interactions from months 3 to 24 were no longer significant (data not shown).
In SPARCL, atorvastatin 80 mg slightly but significantly increased HDL cholesterol in men, but in women, the increase was smaller and not statistically significant, while no consistent gender differences in HDL cholesterol were seen in TNT and IDEAL or CARDS, ASPEN, and ACOT-LLA. No consistent gender differences were seen for the triglyceride-lowering effect of either dose of atorvastatin.
For major CV events and nonfatal myocardial infarction, there was a significant trend across LDL and HDL cholesterol quintiles for men and women. The data for major CV events for LDL and HDL cholesterol quintiles are listed in Table 3 . For HDL cholesterol quintiles, the trend-by-gender interaction was significant (p = 0.0089), indicating that HDL cholesterol was a stronger predictor of major CV events in women than in men. For CHD death, CV death, and all-cause mortality, there was a significant trend across LDL cholesterol quintiles for men but not women; in contrast, for stroke and non-CV death, we observed the opposite effect; namely, the observed trend was significant for women (for non-CV death, the trend was in favor of higher quintiles) and not men.
| Quintile 1 | Quintile 2 | Quintile 3 | Quintile 4 | Quintile 5 | P for trend | |
|---|---|---|---|---|---|---|
| Women: LDL-C (mg/dl) | <65 | 65-83 | 84-103 | 104-128 | >128 | |
| Women: event rate | 132/1727 (7.64%) | 120/1702 (7.05%) | 147/1685 (8.72%) | 136/1694 (8.03%) | 177/1715 (10.32%) | 0.009 |
| Men: LDL-C (mg/dl) | <66 | 66-81 | 82-99 | 100-123 | >123 | |
| Men: event rate | 469/5430 (8.64%) | 512/5807 (8.82%) | 519/5574 (9.31%) | 592/5598 (10.58%) | 554/5603 (9.89%) | <0.0001 |
| Women: HDL-C (mg/dl) | <43 | 43-50 | 51-57 | 58-65 | >65 | |
| Women: event rate | 186/1792 (10.38%) | 162/1613 (10.04%) | 137/1809 (7.57%) | 128/1658 (7.72%) | 108/1760 (6.14%) | <0.0001 |
| Men: HDL-C (mg/dl) | <38 | 38-42 | 43-47 | 48-54 | >54 | |
| Men: event rate | 604/5549 (10.88%) | 635/5905 (10.75%) | 455/5667 (8.03%) | 492/5674 (8.67%) | 485/5734 (8.46%) | <0.0001 |
The unadjusted HRs comparing quintile 5 with quintile 1 in women and men for different end points are listed in Table 4 . Adjusting for diabetes, HDL cholesterol levels, hypertension and smoking increased the HR for major CV events for women from 1.69 to 2.10 (95% CI 1.49 to 2.94) and for men from 1.34 to 1.48 (95% CI 1.22 to 1.79). After adjustment, the trend-by-gender interaction for CHD death was no longer statistically significant (p = 0.56).
