Abstract
Background
Cardiovascular events occur among statin-treated patients, albeit at lower rates. Risk factors for this “residual risk” have not been studied comprehensively. We aimed to identify determinants of this risk above and beyond lipid-related risk factors.
Methods and Results
9,251 coronary patients with LDL cholesterol<130 mg/dL randomized to double-blind atorvastatin 10 or 80 mg/day in the Treating to New Targets (TNT) study had complete on-treatment 1-year lipid data. Median follow-up was 4.9 years. The primary endpoint was major cardiovascular events (n=729): coronary death, non-fatal myocardial infarction, resuscitation after cardiac arrest, or fatal or non-fatal stroke. Multivariable determinants of increased risk were older age (adjusted hazard ratio 1.13 per 1-SD [8.8 years], 95% CI 1.04–1.23), increased body-mass index (BMI) (1.09, 1.02–1.17 per 4.5 kg/m2), male gender (1.33, 1.07–1.65), hypertension (1.38, 1.17–1.63), diabetes (1.33, 1.11–1.60), baseline apolipoprotein B (1.19, 1.11–1.28 per 19 mg/dL) and blood urea nitrogen (1.10, 1.03–1.17 per 4.9 mg/dL), in addition to current smoking, prior cardiovascular disease, and calcium channel blocker use. Determinants of decreased risk were high-dose statin (0.82, 0.70–0.94), aspirin use (0.67, 0.56–0.81), and baseline apolipoprotein A-I (0.91, 0.84–0.99 per 25 mg/dL). On-treatment 1-year lipids or apolipoproteins were not additionally associated with risk in multivariable models. Known baseline variables performed moderately well in discriminating future cases from non-cases (Harrell’s c-index=0.679).
Conclusions
Determinants of residual risk in statin-treated secondary prevention patients included lipid-related and non-lipid factors such as baseline apolipoproteins, increased BMI, smoking, hypertension, and diabetes. A multi-faceted prevention approach should be underscored to address this risk.
Clinical Trial Registration Information
http://clinicaltrials.gov; Unique identifier: NCT00327691
Keywords: apolipoproteins, risk factors, risk prediction, secondary prevention
Statins are the most widely used lipid-lowering agents and the standard of care for individuals with or at high-risk for cardiovascular disease.1, 2 It is currently estimated that 1 out of every 8 US adults are treated with lipid lowering therapy, mostly statins.3 This number is expected to increase, since a large proportion of US adults remain untreated despite meeting guideline recommendations for therapy, including almost two-thirds of individuals at high cardiovascular risk.3
Current guidelines focus on low-density-lipoprotein (LDL) cholesterol lowering as the primary target of therapy, tailoring the level of optimal LDL cholesterol reduction to the individual’s level of risk.1 However, cardiovascular risk among statin-treated individuals remains high and has been termed “residual risk”. Results from a meta-analysis of statin trials involving 90,056 individuals found that the rate of a major cardiovascular event (MCVE) occurring during 5 years of follow-up among statin-treated patients was 21.7% (1 in 5) for individuals with prior cardiovascular disease and 9.5% (1 in 10) for those without prior disease.4, 5 Even after achieving low LDL cholesterol (70–100 mg/dL), residual risk was reduced but remained high in the Treating to New Targets (TNT) trial, with 8.7% of the group allocated to atorvastatin 80 mg experiencing a major event over a 5-year period.6 Thus, there are factors other than LDL cholesterol lowering that determine risk.
The mechanisms underlying this residual risk are uncertain. Identification of these factors is important for more effective tailoring of risk reduction strategies to match the individual level of risk and for development of new treatment targets. This analysis of the TNT study aimed to identify clinical determinants of residual risk by examining the effect of various clinical and lipid-related risk factors among patients with stable coronary disease treated to low LDL cholesterol targets. A secondary aim was to evaluate differences in residual risk according to high- versus low-dose statin therapy.
Methods
Study population
The TNT design has been previously published.6, 7 TNT was a multi-center clinical trial that randomized 10,001 men and women aged 35 to 75 years with stable coronary disease (previous myocardial infarction, previous or present angina with objective evidence of atherosclerotic coronary disease, or previous coronary revascularization procedure) in a parallel-group double-blinded treatment with atorvastatin 80 or 10 mg/day.6, 7 At the screening visit, previously prescribed lipid-lowering drugs were discontinued for at least 6 weeks prior to an open-label 8 week run-in period with atorvastatin 10 mg daily. Patients with triglycerides >600 mg/dl (6.8 mmol/L) after discontinuation of previous lipid-lowering therapy were excluded. At the end of an open-label 8 week run-in with atorvastatin 10 mg/day, patients with mean LDL cholesterol<130 mg/dL (3.4 mmol/L) were randomized to double-blind therapy with atorvastatin 80 or 10 mg/day. In addition to the inclusion and exclusion criteria of the TNT trial as previously published,6 individuals were eligible for this analysis if they had complete 1-year lipid and apolipoprotein measurements and did not experience a MCVE event or death in year 1, resulting in 9,251 individuals for this analysis.
Assessment of clinical and laboratory factors
At the screening visit, an informed consent was signed, demographic characteristics were assessed, vital signs were measured, all concomitant medications were documented, and a medical history was recorded. Metabolic syndrome was defined according to the National Cholesterol Education Program’s 3rd Adult Treatment Panel (ATP-III) definition.8 A body-mass index (BMI) of 28 kg/m2 was used instead of waist circumference which was not assessed at screening.
Baseline and 1-year measurements were performed on fasting blood samples as previously described.2, 9 All baseline measurements were performed on 10 mg of atorvastatin at the end of an 8-week open label run-in. Concentrations of total cholesterol, HDL cholesterol, and triglycerides were quantified by standard enzymatic techniques. LDL cholesterol was calculated with the Friedewald formula when triglycerides were <400 mg/dL.10 For triglycerides ≥400 mg/dL, LDL cholesterol was measured by ultracentrifugation. Concentrations of apolipoprotein B and A-I were measured by immunonephelometry (Behring Nephelometer BNII, Marburg, Germany).2, 9 Other laboratory measurements included white blood cells, uric acid, creatinine, blood urea nitrogen and lactate dehydrogenase.
Assessment of outcomes
Participants were followed for a median of 4.9 years. The primary endpoint was a composite endpoint of MCVE defined as coronary death, nonfatal myocardial infarction, resuscitated cardiac arrest, and fatal or nonfatal stroke. For this analysis, we also examined the expanded secondary endpoint of MCVE plus death, documented angina or arterial revascularization. An independent endpoints committee adjudicated all potential endpoints in a blinded fashion.
Statistical analyses
Summary statistics were calculated according to the occurrence of the primary or secondary endpoint. Candidate variables that could be related to residual risk were selected a priori based on the prior literature and known risk factors. Statistical tests for outcomes were performed according to intention-to-treat on all subjects who survived to year 1 without the endpoint occurring and who also had complete year 1 lipid and apolipoprotein data. Year 1 was chosen because on-treatment lipids and apolipoproteins were part of the major predictor variables in the analysis and this was the first time-point at which a complete panel of lipids and apolipoproteins was measured. The association of each of the candidate variables in relation to the risk of primary (or secondary) endpoint was assessed by univariable Cox proportional hazard models that calculated the hazard ratios (HRs), 95% confidence intervals (CIs), and the χ2 statistics.
Then, also on an a priori basis, we grouped together a set of variables that are the traditional risk factors of age, gender, smoking, diabetes, and hypertension (family history was not obtained in TNT). We included this set of variables, together with randomized treatment assignment in a single model, referring to this model as the “basic model.” Next, variables that were statistically significant at a p-value of <0.10 from the univariable analysis were then entered into the multivariable model. Using an unbiased statistical method of forward stepwise regression, variables that met a critical p-value of <0.05 were selected into the multivariable model.
We then conducted two sensitivity analyses. First, we performed the multivariable model after excluding baseline medication use other the randomization treatment assignment. Second, we expanded the basic model to include the on-treatment 1-year lipids and apolipoproteins in order to identify predictors that remained statistically significant independent of the achieved levels of on-treatment lipids and apolipoproteins.
Treatment-by-variable interactions were assessed using likelihood ratio χ2 statistic and Wald’s p-values to compare models with and without the interaction term. Interactions between the levels of on-treatment lipids and apolipoproteins with the multivariable predictors were also assessed.
Finally, we evaluated the multivariable model discrimination using Harrell’s c-index, which estimates the probability that, of two randomly chosen patients, the patient with the higher prognostic score will be more likely to be a case compared with the patient with the lower prognostic score. 11 Values of c-index near 0.5 indicate that the prognostic score is no better than a coin-flip, while values near 1.0 indicate that the model variables virtually always determine which patient has a better prognosis.
Results
During a median follow-up of 4.9 years, a total of 729 (7.8%) MCVE primary and 1870 secondary endpoints occurred after year 1. Table 1 shows the baseline characteristics for MCVE incident cases and non-cases, which were similar to the overall TNT study population.6 The prevalence of cardiovascular risk factors and prior cardiovascular disease was typical of a population with coronary disease, with a worse cardiovascular risk profile among cases compared with non-cases. Similar results were found when cases were categorized according to the expanded secondary endpoint (results not shown).
Table 1.
Baseline characteristics of cases and non-cases of major cardiovascular events (MCVE) during 4.9 year follow up (N = 9,251)*
| Major cardiovascular events | ||
|---|---|---|
| Cases (n=729) | Non-cases (n=8522) | |
| Treatment: Atorvastatin 80 mg, % | 45.1 | 50.4 |
| Age, years | 62.2 (8.9) | 60.9 (8.8) |
| Height, cm | 173 (9) | 172 (9) |
| Weight, kg | 87.2 (16.9) | 84.5 (15.3) |
| BMI, kg/m2 | 29.3 (5.0) | 28.5 (4.5) |
| Blood pressure, mmHg | ||
| Systolic | 132.9 (17.7) | 130.4 (16.6) |
| Diastolic | 78.3 (9.9) | 77.9 (9.4) |
| Men, % | 83.4 | 80.9 |
| White, % | 93.3 | 94.4 |
| Cardiovascular history, % | ||
| Smoking | ||
| Current | 18.1 | 12.5 |
| Previous | 61.3 | 63.7 |
| Never | 20.6 | 23.8 |
| Diabetes | 23.2 | 14.0 |
| Hypertension | 66.3 | 52.7 |
| Myocardial infarction | 67.9 | 57.2 |
| Angina | 85.5 | 81.1 |
| Cerebrovascular disease | 11.1 | 4.5 |
| Peripheral vascular disease | 20.4 | 10.9 |
| Congestive heart failure | 15.5 | 6.9 |
| Arrhythmia | 23.7 | 17.8 |
| Coronary revascularization | ||
| Angioplasty | 52.3 | 54.2 |
| Bypass | 55.3 | 45.8 |
| Metabolic syndrome, % | 64.3 | 54.8 |
| Chronic kidney disease, % | 38.9 | 31.5 |
| Baseline lipids and apolipoproteins, mg/dL: Mean (SD) | ||
| Total cholesterol | 176.9 (24.6) | 174.3 (23.6) |
| LDL cholesterol | 100.2 (17.5) | 97.1 (17.4) |
| HDL cholesterol | 45.4 (10.1) | 47.6 (11.0) |
| Triglycerides | 158.0 (77.4) | 148.9 (69.0) |
| Median [25th, 75th percentiles] | 138.5 [103.5,194.3] | 133.5 [101.0,181.0] |
| Apolipoprotein A-I | 142.9 (23.1) | 146.5 (24.9) |
| Apolipoprotein B | 115.1 (20.2) | 110.5 (19.0) |
| Year 1 lipids and apolipoproteins, mg/dL: Mean (SD) | ||
| Total cholesterol | 166.8 (34.6) | 162.2 (32.3) |
| LDL cholesterol | 91.8 (27.1) | 87.7 (25.5) |
| HDL cholesterol | 44.6 (10.5) | 46.3 (11.1) |
| Triglycerides | 153.3 (82.2) | 142.2 (81.8) |
| Median [25th, 75th percentiles] | 132.0 [95.0, 192.0] | 123.0 [90.0, 171.0] |
| Apolipoprotein A-I | 140.7 (24.7) | 144.4 (24.8) |
| Apolipoprotein B | 106.5 (25.6) | 101.7 (23.9) |
| Concomitant medication, % | ||
| ACE inhibitor | 36.4 | 26.4 |
| Angiotensin II receptor blockers | 6.6 | 5.0 |
| Beta blockers | 55.6 | 53.8 |
| Calcium channel blockers | 34.3 | 26.0 |
| Aspirin | 80.5 | 87.3 |
| Coumadin or other antiplatlets | 4.5 | 2.9 |
| Prior use of statins | 60.6 | 62.7 |
| Other laboratory parameters: Mean (SD) | ||
| White blood cells (103/mm3) | 6.57 (1.64) | 6.27 (1.62) |
| Glucose (mg/dL) | 113.4 (36.7) | 106.9 (29.4) |
| Uric acid (mmol/L) | 6.43 (1.50) | 6.17 (1.38) |
| Blood urea nitrogen (mg/dL) | 17.77 (5.74) | 16.79 (4.77) |
| Lactate dehydrogenase (mg/dL) | 83.5 (17.5) | 81.6 (17.3) |
The analysis includes only subjects who survived to the year 1 visit without a MCVE and with complete year 1 lipids data since year 1 lipids data were included in the analysis model as predictor variables in all of the subsequent analyses. SD= standard deviation.
Univariable determinants of risk
The primary endpoint of MCVE that occurred after year 1 was reduced with atorvastatin 80 mg versus 10 mg by 18% (P=0.007; Table 2), similar to the overall TNT relative risk reduction of 22%.6 Associations for each of the clinical risk factors, baseline and 1-year laboratory parameters in relation to the primary endpoint of MCVE were obtained from separate Cox regression models (Table 2). Most of the clinical risk factors, and all the baseline and 1-year lipids and apolipoproteins, in addition to baseline white blood cell count, glucose, uric acid, blood urea nitrogen, and lactate dehydrogenase were each statistically significantly associated with MCVE. Similar associations were also noted in relation to the expanded secondary endpoint that included all-cause death, angina, and revascularization (data not shown).
Table 2.
Univariable determinants of post year 1 major cardiovascular events (MCVE)
| Risk factors | SD or % | Major cardiovascular events | P-value* | |
|---|---|---|---|---|
| Hazard Ratio (95% CI) * | χ2 Statistics* | |||
| Treatment: Atorvastatin 80 mg | 50.0% | 0.82 (0.71, 0.95) | 7.22 | 0.007 |
| Age | 8.8 years | 1.15 (1.07, 1.23) | 13.92 | <0.001 |
| Height | 9 cm | 1.04 (0.97, 1.12) | 1.20 | 0.27 |
| Weight | 15.4 kg | 1.18 (1.10, 1.26) | 21.89 | <0.001 |
| BMI | 4.5 kg/m2 | 1.17 (1.10, 1.25) | 23.02 | <0.001 |
| Blood pressure | ||||
| Systolic | 16.7 mmHg | 1.15 (1.07, 1.24) | 15.31 | <0.001 |
| Diastolic | 9.5 mm Hg | 1.03 (0.96, 1.11) | 0.67 | 0.41 |
| Men | 81.1% | 1.18 (0.97, 1.43) | 2.64 | 0.10 |
| White | 94.3% | 0.82 (0.62, 1.10) | 1.75 | 0.19 |
| Cardiovascular history | ||||
| Smoking | ||||
| Current | 12.9% | 1.65 (1.31, 2.08) | 17.53 | <0.001 |
| Previous | 63.5% | 1.11 (0.93, 1.34) | 1.31 | 0.25 |
| Never | 23.6% | 1.00 | ------ | ------ |
| Diabetes | 14.7% | 1.82 (1.54, 2.17) | 46.85 | <0.001 |
| Hypertension | 53.8% | 1.74 (1.49, 2.02) | 49.52 | <0.001 |
| Myocardial infarction | 58.0% | 1.57 (1.34, 1.83) | 32.14 | <0.001 |
| Angina | 81.6% | 1.34 (1.09, 1.65) | 7.71 | 0.006 |
| Cerebrovascular disease | 5.0% | 2.53 (2.01, 3.18) | 61.80 | <0.001 |
| Peripheral vascular disease | 11.7% | 2.03 (1.70, 2.44) | 59.76 | <0.001 |
| Congestive heart failure | 7.5% | 2.40 (1.96, 2.93) | 73.23 | <0.001 |
| Arrhythmia | 18.2% | 1.43 (1.21, 1.70) | 16.94 | <0.001 |
| Coronary revascularization: | ||||
| Angioplasty | 54.1% | 0.92 (0.80, 1.07) | 1.25 | 0.26 |
| Bypass | 46.6% | 1.46 (1.26, 1.69) | 25.93 | <0.001 |
| Metabolic syndrome | 55.6% | 1.47 (1.26, 1.71) | 24.81 | <0.001 |
| Chronic kidney disease | 32.1% | 1.37 (1.18, 1.59) | 17.22 | <0.001 |
| Baseline lipids and apolipoproteins: | ||||
| Total cholesterol | 23.7 mg/dL | 1.10 (1.03, 1.18) | 7.17 | 0.007 |
| LDL cholesterol | 17.5 mg/dL | 1.17 (1.09, 1.26) | 18.37 | <0.001 |
| HDL cholesterol | 47.4 mg/dL | 0.81 (0.75, 0.88) | 25.85 | <0.001 |
| Triglycerides | 69.7 mg/dL | 1.12 (1.05, 1.20) | 12.16 | <0.001 |
| Apolipoprotein A-I | 24.8 mg/dL | 0.86 (0.79, 0.92) | 16.04 | <0.001 |
| Apolipoprotein B | 19.1 mg/dL | 1.24 (1.16, 1.33) | 36.22 | <0.001 |
| Year 1 lipids and apolipoproteins: | ||||
| Total cholesterol | 32.5 mg/dL | 1.13 (1.06, 1.22) | 12.56 | < 0.001 |
| LDL cholesterol | 25.7 mg/dL | 1.16 (1.08, 1.24) | 16.93 | <0.001 |
| HDL cholesterol | 11.1 mg/dL | 0.85 (0.79, 0.92) | 16.62 | <0.001 |
| Triglycerides | 81.9 mg/dL | 1.10 (1.05, 1.16) | 13.12 | <0.001 |
| Apolipoprotein A-I | 24.1 mg/dL | 0.86 (0.80, 0.93) | 15.55 | <0.001 |
| Apolipoprotein B | 24.1 mg/dL | 1.20 (1.12, 1.28) | 26.43 | <0.001 |
| Concomitant medication | ||||
| ACE inhibitor | 27.2% | 1.57 (1.35, 1.83) | 34.45 | <0.001 |
| Angiotensin II receptor blockers | 5.2% | 1.31 (0.98, 1.76) | 3.27 | 0.07 |
| Beta blockers | 53.9% | 1.07 (0.93. 1.24) | 0.91 | 0.34 |
| Calcium channel blockers | 26.7% | 1.46 (1.25, 1.70) | 23.52 | <0.001 |
| Aspirin | 86.8% | 0.62 (0.51, 0.74) | 26.88 | <0.001 |
| Coumadin or other antiplatlets | 3.0% | 1.55 (1.09, 2.19) | 5.96 | 0.01 |
| Prior use of statins | 62.6% | 0.93 (0.80, 1.08) | 0.87 | 0.35 |
| Other laboratory parameters: | ||||
| White blood cells (natural log) | 0.25 103/mm3 | 1.20 (1.12, 1.29) | 24.55 | <0.001 |
| Glucose | 30.1 mg/dL | 1.18 (1.11, 1.24) | 32.32 | <0.001 |
| Uric acid | 1.39 mmol/L | 1.18 (1.10, 1.27) | 22.38 | <0.001 |
| Blood urea nitrogen | 4.86 mg/dL | 1.17 (1.11, 1.24) | 29.30 | <0.001 |
| Lactate dehydrogenase | 17.3 mg/dL | 1.10 (1.03, 1.16) | 9.02 | 0.003 |
Hazard ratios, 95% confidence intervals (CIs), and p-values are based on unviariable Cox proportional hazard analysis including the corresponding variable in the model. Hazard ratios and the corresponding 95% CIs are calculated based on 1 standard deviation (SD) increase for the continuous variables.
Multivariable determinants of risk
Using forward stepwise regression, statistically significant variables were added to the basic model (randomization treatment, age, gender, smoking, diabetes, and hypertension). Multivariable determinants of increased risk (Table 3) were older age, increased BMI, male gender, hypertension, diabetes, baseline apolipoprotein B and blood urea nitrogen, current smoking, prior cardiovascular disease, and calcium channel blocker use. Determinants of decreased risk were high-dose statin, aspirin use, and baseline apolipoprotein A-I. On-treatment 1-year levels of lipids and apolipoproteins were not selected into the multivariable model because they were not associated with risk after taking into account baseline apolipoproteins and clinical risk factors. When we repeated the multivariable analysis in Table 3 for the secondary expanded endpoint, nearly identical results were obtained, except that both baseline apolipoprotein B and 1-year apolipoprotein B and A-I were also statistically significant (data not shown).
Table 3.
Multivariable determinants of post year 1 major cardiovascular events (MCVE)
| Risk factors | Major cardiovascular events | χ2 Statistics* | P- value* |
|---|---|---|---|
| Hazard Ratio (95% CI) * | |||
| Basic model | |||
| Treatment (Atorvastatin 80 mg) | 0.82 (0.70, 0.94) | 7.43 | 0.006 |
| Age (years): per 1SD | 1.13 (1.04, 1.23) | 8.04 | 0.005 |
| Men | 1.33 (1.07, 1.65) | 6.69 | 0.01 |
| Smoking | |||
| Current | 1.65 (1.28, 2.12) | 15.35 | <0.001 |
| Previous | 1.00 (0.82, 1.20) | 0.01 | 0.96 |
| Never | 1.00 | ------ | ------ |
| Hypertension | 1.38 (1.17, 1.63) | 14.51 | <0.001 |
| Diabetes | 1.33 (1.11, 1.60) | 9.48 | 0.002 |
| Variables selected by forward stepwise elimination | |||
| Body mass index(kg/m2): per 1SD | 1.09 (1.02, 1.17) | 6.34 | 0.01 |
| Myocardial infarction | 1.60 (1.36, 1.87) | 32.79 | <0.001 |
| Angina | 1.36 (1.10, 1.68) | 8.16 | 0.004 |
| Cerebrovascular disease | 1.73 (1.36, 2.19) | 19.80 | <0.001 |
| Peripheral vascular disease | 1.32 (1.09, 1.61) | 7.84 | 0.005 |
| Congestive heart failure | 1.54 (1.24, 1.90) | 15.59 | <0.001 |
| Coronary revascularization: Bypass | 1.26 (1.08, 1.47) | 8.44 | 0.004 |
| Baseline apolipoprotein A-I (mg/dL): per 1SD | 0.91 (0.84, 0.99) | 5.23 | 0.02 |
| Baseline apolipoprotein B (mg/dL): per 1SD | 1.19 (1.11, 1.28) | 21.87 | <0.001 |
| Use of calcium channel blockers | 1.31 (1.12, 1.54) | 10.99 | <0.001 |
| Use of aspirin | 0.67 (0.56, 0.81) | 17.31 | <0.001 |
| Baseline blood urea nitrogen (mg/dL): per 1SD | 1.10 (1.03, 1.17) | 8.92 | 0.003 |
Hazard ratios, 95% confidence intervals (CIs), and p-values are based on multivariable Cox proportional hazard analysis forcing age, gender, smoking, hypertension, and diabetes in the model; and the remaining significant (p< 0.10) variables identified from the univariable analyses were entered into the model by forward stepwise selection process with a critical value of 0.05. Hazard ratios and the corresponding 95% CIs are calculated based on 1 standard deviation (SD) increase for the continuous variables.
Sensitivity analyses
We performed 2 additional sensitivity analyses. First, we repeated the multivariable analysis in Table 3 after excluding baseline concurrent medication use, obtaining very similar results (Supplementary Table).
Then, in order to determine if the multivariable determinants remained associated with risk independent of the achieved levels of on-treatment 1-year lipids and apolipoproteins, we performed a second sensitivity analysis by expanding the basic model to include the 1-year lipids and apolipoproteins. None of the 1-year lipids or apolipoproteins were associated with risk of the primary or secondary endpoint.
Tests for interactions
We tested for whether the randomization treatment of atorvastatin 80 mg versus 10 mg modified the association of the multivariable determinants of risk. Generally similar predictors of risk were seen among subjects allocated atorvastatin 10 mg vs 80 mg (Table 4). There were no statistically significant interactions at a P value of <0.01.
Table 4.
Multivariable determinants of post year 1 major cardiovascular event by randomization treatment groups (atorvastatin 80 mg versus 10 mg)
| Risk factors | Major cardiovascular events
|
||||
|---|---|---|---|---|---|
| Atorvastatin 10 mg (N = 4628) | Atorvastatin 80 mg (N = 4623) | P for interaction† | |||
| Hazard Ratio (95% CI)* | P-value* | Hazard Ratio (95% CI)* | P-value* | ||
| Basic model | |||||
| Age (years): per 1SD | 1.09 (0.98, 1.23) | 0.13 | 1.17 (1.03, 1.33) | 0.01 | 0.68 |
| Men | 1.21 (0.91, 1.62) | 0.19 | 1.52 (1.09, 2.11) | 0.01 | 0.47 |
| Smoking | |||||
| Current | 1.57 (1.19, 2.08) | 0.002 | 1.70 (1.26, 2.31) | <0.001 | 0.74 |
| Previous | 0.95 (0.74, 1.23) | 0.70 | 1.01 (0.81, 1.43) | 0.61 | 0.49 |
| Never | 1.00 | ------ | 1.00 | ------ | |
| Hypertension | 1.30 (1.04, 1.62) | 0.02 | 1.47 (1.15, 1.89) | 0.002 | 0.44 |
| Diabetes | 1.35 (1.05, 1.74) | 0.02 | 1.29 (0.99, 1.69) | 0.06 | 0.84 |
| Variables selected by forward stepwise selection | |||||
| Body mass index (kg/m2): per 1SD | 1.06 (0.97, 1.16) | 0.22 | 1.15 (1.03, 1.28) | 0.01 | 0.38 |
| Myocardial infarction | 1.43 (1.16, 1.76) | <0.001 | 1.83 (1.43, 2.35) | <0.001 | 0.17 |
| Angina | 1.39 (1.04, 1.85) | 0.02 | 1.32 (0.97, 1.81) | 0.08 | 0.70 |
| Cerebrovascular disease | 1.78 (1.29, 2.45) | <0.001 | 1.68 (1.17, 2.41) | 0.005 | 0.79 |
| Peripheral vascular disease | 1.21 (0.92, 1.59) | 0.17 | 1.48 (1.12, 1.97) | 0.007 | 0.30 |
| Congestive heart failure | 1.29 (0.95, 1.75) | 0.10 | 1.83 (1.36, 2.48) | <0.001 | 0.09 |
| Coronary revascularization: Bypass | 1.35 (1.10, 1.67) | 0.005 | 1.15 (0.91, 1.44) | 0.25 | 0.31 |
| Baseline apolipoprotein A-I (mg/dL):per 1SD | 0.88 (0.79, 0.99) | 0.03 | 0.94 (0.83, 1.06) | 0.33 | 0.67 |
| Baseline apolipoprotein B (mg/dL) :per 1SD | 1.25 (1.14, 1.38) | <0.001 | 1.12 (1.00, 1.25) | 0.05 | 0.05 |
| Use of calcium channel blockers | 1.52 (1.23, 1.89) | <0.001 | 1.08 (0.85, 1.38) | 0.53 | 0.04 |
| Use of aspirin | 0.73 (0.56, 0.95) | 0.02 | 0.61 (0.46, 0.79) | <0.001 | 0.25 |
| Baseline blood urea nitrogen (mg/dL): per 1SD | 1.11 (1.02, 1.22) | 0.02 | 1.08 (0.93, 1.18) | 0.07 | 0.63 |
Hazard ratio, 95% confidence intervals (CIs), and p-values are based on multivariable Cox proportional hazard analysis including all specified variables in the model. Hazard ratios and the corresponding 95% CIs are calculated based on 1 standard deviation (SD) increase for the continuous variables.
The p-values for the interaction terms between treatment and the corresponding variables are based on the Wald’s test of the corresponding interaction terms in the multivariable model listed above.
Since individuals who achieve lower lipid targets on therapy still experience events, we determined whether the multivariable determinants differed among patients based on their achieved levels of 1-year lipids and apolipoproteins. There were no statistically significant interactions at a P value of <0.01.
Prognostic model discrimination
Finally, we assessed the overall model discrimination for the multivariable model using Harrell’s c-index,11 a generalization of the area under the receiver operator characteristic curve. The multivariable model (Table 3) provided acceptable discrimination of cases from non-cases, with a c-index of 0.679. A similar multivariable model that excluded baseline medication use (Supplementary Table) resulted in a slightly lower c-index of 0.673.
Discussion
In the TNT trial of secondary prevention patients with low LDL cholesterol and triglycerides less than 600 mg/dL, residual risk was multi-factorial and related to baseline lipid-related and non-lipid risk factors, including baseline apolipoproteins, increased BMI, smoking, hypertension, and diabetes. Generally similar determinants of risk were identified among subjects allocated to high-versus low-dose statin therapy. Known baseline clinical and lipid-related variables performed moderately well in discriminating cases from non-cases. Thus, a multi-faceted secondary prevention approach emphasizing modifiable traditional risk factors should be underscored in order to reduce residual risk.
It is commonly believed that residual risk on statin treatment is related to the achieved levels of high-density-lipoprotein (HDL) cholesterol or triglycerides, in addition to achieved LDL cholesterol.12 At first glance, the findings of this TNT analysis may seem discordant with the results of a prior TNT analysis that found an inverse association of HDL cholesterol with MCVE. But in the prior analysis, there was no association between HDL cholesterol and MCVE (p-value=0.45) after adjustment for baseline apolipoproteins.13 The current study also found univariable associations for HDL cholesterol and triglycerides with risk that became non-statistically significant in multivariable models that additionally adjusted for baseline apolipoproteins and other risk factors. In contrast, among primary prevention patients with low LDL cholesterol but elevated C-reactive protein in the Justification for the Use of statins in Prevention: an Intervention Trial Evaluating Rosuvastatin (JUPITER) trial, neither on-treatment HDL cholesterol nor triglycerides were associated with residual risk, even without adjusting for apolipoproteins.14, 15 The JUPITER study population differed from TNT in being a primary prevention non-diabetic population recruited on the basis of chronic inflammation, and these differences in the study populations may contribute to differences in prognostic markers. However, several other trials also did not find statistically significant associations for on-treatment HDL cholesterol or triglycerides with residual risk among the active or more intensely treated arms.16–18
There is limited prior data evaluating the magnitude of residual risk that is explained by clinical risk factors. However, 2 prior studies from hypercholesterolemic primary prevention populations identified similar risk factors for residual risk as were found in the current TNT study of secondary prevention. In a previous study from the West of Scotland Coronary Prevention Study (WOSCOPS), which only included middle-aged hypercholesterolemic men without prior myocardial infarction, important predictors of risk included older age, smoking, blood pressure, diabetes, and baseline total/HDL cholesterol ratio, and a 5-year risk score for residual risk was modeled.19, 20 In a subsequent analysis from hypercholesterolemic asymptomatic men and women from the Management of Elevated Cholesterol in the Primary Prevention Group of Adult Japanese (MEGA) study, the risk factors for incident coronary events were male sex, older age, diabetes, hypertension, smoking, low baseline HDL cholesterol, and increased BMI (women only).21
Furthermore, the current findings do not support the routine re-measurement of on-treatment 1-year lipids or apolipoproteins once a baseline measurement has been obtained on statin therapy, because 1-year levels were not related to residual risk above and beyond the initial (on-treatment) values and clinical risk factors. A prior TNT analysis found positive association for on-treatment apolipoproteins and non-HDL cholesterol with residual risk in minimally-adjusted models (age- and sex-adjusted). 2 The current study extends the prior findings by demonstrating that once additional clinical risk factors and baseline levels of apolipoproteins are included in more comprehensive multivariable models, there is no independent association for 1-year on-treatment levels of apolipoproteins with residual risk. Notably, in both MEGA and WOSCOPS, on-treatment lipid values were not associated with risk in multivariable models.20, 21
Risk factors for CVD often cluster such that individuals with dyslipidemia may also have a number of cardiovascular risk factors.22,23 Even after achieving low LDL cholesterol levels, as in TNT, a greater duration and/or burden of atherosclerotic disease are often present among coronary patients that increase their residual risk. For example, the increased risk associated with calcium channel blocker use is at least in part related to “confounding by indication”, since patients taking calcium channel blockers at baseline probably had a higher burden (severity and/or frequency) of angina. For the rest of the multivariable determinants of risk, we believe that our results show the expected (and not paradoxical) relationships of established risk factors with clinical outcomes in the secondary prevention setting because of 1) better assessment of potential confounders in the TNT trial and 2) less differential surveillance bias. Detailed risk factor data and information about potential confounders were well-assessed in the TNT clinical trial, and hence the results are less prone to the confounder bias that may be seen in other prospective studies that did not account sufficiently for potential confounders and which likely resulted in the “paradox phenomenon”.24 We also believe that there was less differential surveillance bias in the TNT study, since it was a randomized double-blinded clinical trial, as compared with unblinded prospective studies that may be more prone to this type of bias.
The multivariable model that included these clinical risk factors, concurrent medication use, and prior cardiovascular history resulted in acceptable discrimination of future cases and non-cases. Studies are needed to identify additional factors that may contribute to residual risk, and to test the comparative effectiveness of treatment strategies that simultaneously target multiple risk factors.
The present study has potential limitations and several strengths. TNT excluded patients with LDL cholesterol >130 mg/dL on atorvastatin 10 mg or untreated triglycerides >600 mg/dL. This prospective analysis, while conducted within a clinical trial, was observational and may be subject to bias, since individuals who become cases may differ from non-cases in unmeasured ways. While we examined clinical and traditional risk factor data, novel risk factor data was not available on the entire TNT population for analysis. However, a prior case-control study found little association for a selected group of non-lipid biomarkers with residual risk.25 Finally, it is unclear if our results would be applicable to other individuals from primary or secondary prevention who were excluded from the trial.
Strengths of the study are the large number of individuals with baseline and 1-year lipids and apolipoproteins, as well as the detailed information on clinical risk factors. To our knowledge this is the first analysis of the role of clinical risk factors in a secondary prevention study, in particular one in which baseline LDL cholesterol was low and the achieved LDL cholesterol was even lower with half the study population allocated to potent statin therapy. Finally, because the TNT study design tested a low- and high-dose statin in a randomized fashion, we were able to take advantage of this clinical trial design using the intention-to-treat analysis to test for treatment-by-variable interactions.
Conclusions
In summary, the key finding from this study is that residual risk among statin-allocated coronary patients was multi-factorial and related to baseline lipid-related and non-lipid risk factors, including baseline apolipoproteins, increased BMI, smoking, hypertension, and diabetes. Known baseline clinical and lipid-related variables performed moderately well in discriminating future cases from non-cases. Thus, a multi-faceted secondary prevention approach targeting modifiable risk factors should be underscored as the cornerstone of optimal cardiovascular risk assessment and prevention.
Supplementary Material
Acknowledgments
Funding Sources: The TNT trial was funded by Pfizer, Inc, which had no role in the design of the current study, drafting this report or in the decision to submit these analyses for publication. Dr Mora received research grant support from the National Heart, Lung, and Blood Institute (K08 HL094375).
Footnotes
This manuscript was presented in part as an oral abstract on November 12, 2011 at the Annual Scientific Sessions of the American Heart Association in Orlando, Florida
Conflict of Interest Disclosures: Dr Mora has received research grant support from AstraZeneca and Merck, served as a consultant for Pfizer and Quest Diagnostics, and received non-promotional speaker honorarium from Abbott and AstraZeneca. Dr Wenger has received research grants from Merck, Gilead Sciences (formerly CV Therapeutics), Pfizer, NHLBI, Abbott and Eli Lilly, and has served as a consultant/advisory board member or received honoraria from Astra Zeneca, Gilead Sciences (formerly CV Therapeutics), Abbott, Merck, Pfizer, and Medtronic. Dr DeMicco and Dr Breazna are employees of Pfizer. Dr Boekholdt and Dr Arsenault have served as consultant/advisory board members for Pfizer. Dr Deedwania has received research grant support from Pfizer and served as consultant/advisory board member of Pfizer and Novartis. Dr Kastelein has served as consultant/advisory board member of Pfizer. Dr Waters has served as consultant/advisory board member of Merck Schering-Plough, Genetech, Aegerion, Roche, Sanofi-Aventis, Pfizer, and Anthera.
References
- 1.Executive summary of the third report of the National Cholesterol Education Program (NCEP) expert panel on detection, evaluation, and treatment of high blood cholesterol in adults (Adult Treatment Panel III) JAMA. 2001;285:2486–2497. doi: 10.1001/jama.285.19.2486. [DOI] [PubMed] [Google Scholar]
- 2.Kastelein JJ, van der Steeg WA, Holme I, Gaffney M, Cater NB, Barter P, Deedwania P, Olsson AG, Boekholdt SM, Demicco DA, Szarek M, LaRosa JC, Pedersen TR, Grundy SM. Lipids, apolipoproteins, and their ratios in relation to cardiovascular events with statin treatment. Circulation. 2008;117:3002–3009. doi: 10.1161/CIRCULATIONAHA.107.713438. [DOI] [PubMed] [Google Scholar]
- 3.Kuklina EV, Yoon PW, Keenan NL. Trends in high levels of low-density lipoprotein cholesterol in the United States, 1999–2006. JAMA. 2009;302:2104–2110. doi: 10.1001/jama.2009.1672. [DOI] [PubMed] [Google Scholar]
- 4.Baigent C, Keech A, Kearney PM, Blackwell L, Buck G, Pollicino C, Kirby A, Sourjina T, Peto R, Collins R, Simes R. Efficacy and safety of cholesterol-lowering treatment: Prospective meta-analysis of data from 90,056 participants in 14 randomised trials of statins. Lancet. 2005;366:1267–1278. doi: 10.1016/S0140-6736(05)67394-1. [DOI] [PubMed] [Google Scholar]
- 5.Baigent C, Blackwell L, Emberson J, Holland LE, Reith C, Bhala N, Peto R, Barnes EH, Keech A, Simes J, Collins R. Efficacy and safety of more intensive lowering of LDL cholesterol: A meta-analysis of data from 170,000 participants in 26 randomised trials. Lancet. 2010;376:1670–1681. doi: 10.1016/S0140-6736(10)61350-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.LaRosa JC, Grundy SM, Waters DD, Shear C, Barter P, Fruchart JC, Gotto AM, Greten H, Kastelein JJ, Shepherd J, Wenger NK. Intensive lipid lowering with atorvastatin in patients with stable coronary disease. N Engl J Med. 2005;352:1425–1435. doi: 10.1056/NEJMoa050461. [DOI] [PubMed] [Google Scholar]
- 7.Waters DD, Guyton JR, Herrington DM, McGowan MP, Wenger NK, Shear C. Treating to new targets (TNT) study: Does lowering low-density lipoprotein cholesterol levels below currently recommended guidelines yield incremental clinical benefit? Am J Cardiol. 2004;93:154–158. doi: 10.1016/j.amjcard.2003.09.031. [DOI] [PubMed] [Google Scholar]
- 8.Grundy SM, Cleeman JI, Daniels SR, Donato KA, Eckel RH, Franklin BA, Gordon DJ, Krauss RM, Savage PJ, Smith SC, Jr, Spertus JA, Costa F. Diagnosis and management of the metabolic syndrome: An American Heart Association/National Heart, Lung, and Blood Institute Scientific Statement. Circulation. 2005;112:2735–2752. doi: 10.1161/CIRCULATIONAHA.105.169404. [DOI] [PubMed] [Google Scholar]
- 9.Faergeman O, Holme I, Fayyad R, Bhatia S, Grundy SM, Kastelein JJ, LaRosa JC, Larsen ML, Lindahl C, Olsson AG, Tikkanen MJ, Waters DD, Pedersen TR. Plasma triglycerides and cardiovascular events in the treating to new targets and incremental decrease in end-points through aggressive lipid lowering trials of statins in patients with coronary artery disease. Am J Cardiol. 2009;104:459–463. doi: 10.1016/j.amjcard.2009.04.008. [DOI] [PubMed] [Google Scholar]
- 10.Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem. 1972;18:499–502. [PubMed] [Google Scholar]
- 11.Harrell FE, Jr, Lee KL, Mark DB. Multivariable prognostic models: Issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med. 1996;15:361–387. doi: 10.1002/(SICI)1097-0258(19960229)15:4<361::AID-SIM168>3.0.CO;2-4. [DOI] [PubMed] [Google Scholar]
- 12.Fruchart JC, Sacks F, Hermans MP, Assmann G, Brown WV, Ceska R, Chapman MJ, Dodson PM, Fioretto P, Ginsberg HN, Kadowaki T, Lablanche JM, Marx N, Plutzky J, Reiner Z, Rosenson RS, Staels B, Stock JK, Sy R, Wanner C, Zambon A, Zimmet P. The residual risk reduction initiative: A call to action to reduce residual vascular risk in patients with dyslipidemia. Am J Cardiol. 2008;102:1K–34K. doi: 10.1016/S0002-9149(08)01833-X. [DOI] [PubMed] [Google Scholar]
- 13.Barter P, Gotto AM, LaRosa JC, Maroni J, Szarek M, Grundy SM, Kastelein JJ, Bittner V, Fruchart JC. HDL cholesterol, very low levels of LDL cholesterol, and cardiovascular events. N Engl J Med. 2007;357:1301–1310. doi: 10.1056/NEJMoa064278. [DOI] [PubMed] [Google Scholar]
- 14.Ridker PM, Genest J, Boekholdt SM, Libby P, Gotto AM, Nordestgaard BG, Mora S, MacFadyen JG, Glynn RJ, Kastelein JJ. HDL cholesterol and residual risk of first cardiovascular events after treatment with potent statin therapy: An analysis from the Jupiter trial. Lancet. 2010;376:333–339. doi: 10.1016/S0140-6736(10)60713-1. [DOI] [PubMed] [Google Scholar]
- 15.Mora S, Glynn RJ, Boekholdt M, Nordestgaard BG, Kastelein JJ, Ridker PM. On-treatment non-HDL cholesterol, apolipoprotein B, triglycerides, and lipid ratios in relation to residual vascular risk after treatment with potent statin therapy: The Jupiter trial. J Am Coll Cardiol. doi: 10.1016/j.jacc.2011.12.035. in press. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Pedersen TR, Olsson AG, Faergeman O, Kjekshus J, Wedel H, Berg K, Wilhelmsen L, Haghfelt T, Thorgeirsson G, Pyorala K, Miettinen T, Christophersen B, Tobert JA, Musliner TA, Cook TJ. Lipoprotein changes and reduction in the incidence of major coronary heart disease events in the Scandinavian Simvastatin Survival Study (4S) Circulation. 1998;97:1453–1460. doi: 10.1161/01.cir.97.15.1453. [DOI] [PubMed] [Google Scholar]
- 17.Gotto AM, Jr, Whitney E, Stein EA, Shapiro DR, Clearfield M, Weis S, Jou JY, Langendorfer A, Beere PA, Watson DJ, Downs JR, de Cani JS. Relation between baseline and on-treatment lipid parameters and first acute major coronary events in the Air Force/Texas Coronary Atherosclerosis Prevention Study (AFCAPS/TexCAPS) Circulation. 2000;101:477–484. doi: 10.1161/01.cir.101.5.477. [DOI] [PubMed] [Google Scholar]
- 18.Ray KK, Cannon CP, Cairns R, Morrow DA, Ridker PM, Braunwald E. Prognostic utility of apoB/AI, total cholesterol/HDL, non-HDL cholesterol, or hs-Crp as predictors of clinical risk in patients receiving statin therapy after acute coronary syndromes: Results from PROVE IT-TIMI 22. Arterioscler Thromb Vasc Biol. 2009;29:424–430. doi: 10.1161/ATVBAHA.108.181735. [DOI] [PubMed] [Google Scholar]
- 19.The West of Scotland Coronary Prevention Study Group. Baseline risk factors and their association with outcome in the West of Scotland Coronary Prevention Study. Am J Cardiol. 1997;79:756–762. doi: 10.1016/s0002-9149(96)00863-6. [DOI] [PubMed] [Google Scholar]
- 20.Influence of pravastatin and plasma lipids on clinical events in the West of Scotland Coronary Prevention Study (WOSCOPS) Circulation. 1998;97:1440–1445. doi: 10.1161/01.cir.97.15.1440. [DOI] [PubMed] [Google Scholar]
- 21.Ishikawa T, Mizuno K, Nakaya N, Ohashi Y, Tajima N, Kushiro T, Teramoto T, Uchiyama S, Nakamura H. The relationship between the effect of pravastatin and risk factors for coronary heart disease in Japanese patients with hypercholesterolemia. Circ J. 2008;72:1576–1582. doi: 10.1253/circj.cj-08-0095. [DOI] [PubMed] [Google Scholar]
- 22.Becker DM, Becker LC, Pearson TA, Fintel DJ, Levine DM, Kwiterovich PO. Risk factors in siblings of people with premature coronary heart disease. J Am Coll Cardiol. 1988;12:1273–1280. doi: 10.1016/0735-1097(88)92611-3. [DOI] [PubMed] [Google Scholar]
- 23.Genest JJ, Jr, Martin-Munley SS, McNamara JR, Ordovas JM, Jenner J, Myers RH, Silberman SR, Wilson PW, Salem DN, Schaefer EJ. Familial lipoprotein disorders in patients with premature coronary artery disease. Circulation. 1992;85:2025–2033. doi: 10.1161/01.cir.85.6.2025. [DOI] [PubMed] [Google Scholar]
- 24.Dahabreh IJ, Kent DM. Index event bias as an explanation for the paradoxes of recurrence risk research. JAMA. 2011;305:822–823. doi: 10.1001/jama.2011.163. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Arsenault BJ, Barter P, DeMicco DA, Bao W, Preston GM, LaRosa JC, Grundy SM, Deedwania P, Greten H, Wenger NK, Shepherd J, Waters DD, Kastelein JJP. Prediction of cardiovascular events in statin-treated stable coronary patients by lipid and nonlipid biomarkers. J Am Coll Cardiol. 2011;57:63–69. doi: 10.1016/j.jacc.2010.06.052. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.