Critical Review of High-Sensitivity C-Reactive Protein and Coronary Artery Calcium for the Guidance of Statin Allocation

JUPITER Trial

hsCRP is an acute phase inflammatory protein shown to predict the likelihood of future cardiovascular events regardless of baseline Framingham risk score, low-density lipoprotein-cholesterol (LDL-C) level, or presence of metabolic syndrome.^5,6 The large-scale Women’s Health Study demonstrated that the absolute risk of future cardiovascular events was higher in the group with low levels of LDL-C and high levels of hsCRP than those with high levels of LDL-C but low levels of hsCRP.⁵ Because statins have anti-inflammatory effects as well as cholesterol-lowering effects,⁷ and they are known to reduce hsCRP levels in human populations,⁸ it was theorized that risk assessment based on hsCRP levels might provide a superior prediction of response to statin therapy. This theory was buoyed by the post hoc analyses of Air Force/Texas Coronary Atherosclerosis Prevention (AFCAPS/TexCAPS) trial and the Cholesterol and Recurrent Events (CARE) trial, which showed that lovastatin allocation significantly lowered primary cardiovascular outcomes among subjects with relatively low LDL-C levels (LDL-C<149 mg/dL) and elevated hsCRP levels (hsCRP>1.6 mg/L).^8,9

Thus, the stated rationale of the JUPITER trial was to prove that identification of subclinical inflammation through hsCRP testing is an effective screening method in patients without hyperlipidemia “to detect high-risk individuals for whom the NNT is small enough to make prophylactic statin therapy cost effective”.³

St. Francis Heart Trial

Multiple observational studies have verified that the CAC score is a robust, independent predictor of future cardiovascular events.¹⁰ The amount of calcium detected by noncontrast cardiac computed tomography closely correlates with the total atherosclerosis plaque burden of the coronary arteries.¹¹ Indeed, a finding of zero CAC corresponds to low future event rates,¹² whereas elevated CAC markedly increases risk even among low-risk patients.¹³ Studies have demonstrated that medical treatment with statins can delay coronary artery plaque progression, delipidate and stabilize plaque, and cause regression of existing atherosclerosis.¹⁴ Accordingly, it has been hypothesized that treating individuals with higher CAC scores will result in improved outcomes.

Interpretation of Study Rationales

Both of the above study rationales were based on abundant clinical and laboratory evidence that provided the necessary justification to enroll a patient population defined by the respective biomarker. An ideal biomarker should be able to improve cardiac risk prediction, decrease the NNT with statins, increase statin number needed to harm, and improve statin cost-effectiveness. To this end, both studies share the rationale that a biomarker could select a population with higher relative and absolute risk that benefits most from statin therapy.

The JUPITER trial attempted to evaluate the use of hsCRP in those with normal LDL levels who would not have warranted statin treatment under Adult Treatment Panel III guidelines. In contrast, the St. Francis Heart Study preceded the contemporary more aggressive lipid guidelines, enrolling subjects with higher LDL (mean LDL, 143 mg/dL). Therefore, the former study was an explicit attempt to expand statin indication by the hsCRP-guided approach, whereas the latter was a proof of the concept of CAC-guided statin therapy.

However, because the biomarkers were measured in an unblinded fashion and enrollment occurred based on the result, neither design was capable of completely assessing the ability of either inflammation (as measured by hsCRP) or subclinical calcified atherosclerosis (as measured by CAC) to guide clinical practice or predict relative statin effect as a preventive therapy. This argument will be discussed further below.

JUPITER Trial

The JUPITER trial was financially supported by Astra-Zeneca. The sponsor performed the trial data collection and study site monitoring but was not involved in data analysis or article drafting.^15,16

St. Francis Heart Trial

The St. Francis Heart trial was mainly supported by the St. Francis Hospital Foundation. Parke-Davis Pharmaceuticals, which is a subsidiary of Pfizer, Inc, supplied the statin and placebo at no cost.^4,17

Interpretation of Funding Source

Both of the trials were investigator-initiated, but the JUPITER trial had more involvement by the pharmaceutical company in the study process, from patient enrollment to data accumulation.¹⁸ Although it is well known that industrial-funded trials tend to show more positive findings than trials funded by nonprofit organizations, there is no evidence that the involvement of the pharmaceutical company affected the study outcome.¹⁹

JUPITER Trial

The JUPITER trial was a randomized, double-blinded, placebo-controlled, multicenter trial conducted in 26 countries, of which the United States, United Kingdom, Canada, and South Africa were the main sites. After initial screening, a study population with LDL-C <130 mg/dL and hsCRP ≥2 mg/L were randomized to 2 groups. One group received rosuvastatin 20 mg daily and the other group received a placebo.

St. Francis Heart Trial

The St. Francis Heart trial was a randomized, double-blinded, placebo-controlled, single center trial that was coordinated at St. Francis Hospital, New York, United States. After the initial computed tomography screening, subjects with CAC scores >80th percentile for age and sex were identified and randomly assigned to either an atorvastatin 20 mg daily (in addition to vitamins C and E) or a placebo.

Interpretation of Study Designs

The JUPITER trial exclusively enrolled patients with elevated hsCRP >2 mg/L and did not have an arm for subjects with LDL-C <130 mg/dL and hsCRP <2 mg/L. The JUPITER investigators decided not to include this arm based on the results of post hoc analyses from the AFCAPS/TexCAPS study and the CARE trial, which demonstrated that people with low hsCRP levels had no cardiovascular benefit with statin treatment.⁹

However, it is still debatable whether statin therapy truly has no or little benefit in subjects with low hsCRP. A post hoc analysis from the Heart Protection Study demonstrated that there was a similar 29% relative risk reduction in major cardiovascular events even in people with low hsCRP levels (<1.25 mg/L) when they were treated with simvastatin 40 mg daily.²⁰ Furthermore, reanalysis of the AFCAPS/TexCAPS trial for a true treatment–subgroup interaction did not show true effect modification by hsCRP status (interaction P value=0.305).²¹

Evaluation of a negative biomarker arm is essential to assess the effectiveness of a biomarker-guided statin allocation strategy, because otherwise it is impossible to disprove that statin treatment is effective for all subjects, irrespective of the level of hsCRP.²² Therefore, the JUPITER trial is best considered a well-conducted cholesterol-lowering trial rather than a biomarker trial because the efficacy of hsCRP as a biomarker was not fully assessed (Figure 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Study designs for assessing the efficacy of statins, eg, the JUPITER trial (A), and a true biomarker trial (B).

Like the JUPITER trial, the St. Francis Heart trial does not qualify as a biomarker trial. The St. Francis Heart trial lacks an arm with CAC scores <80th percentile for age and sex, which makes it difficult to exclude the possibility that statins provide equivalent benefit in patients with low CAC scores.

JUPITER Trial

A total of 89 890 people were screened for enrollment and 17 802 were included and ultimately randomly assigned to a study group. The main excluding factors were LDL-C levels ≥130 mg/dL (52.2%) and hsCRP levels <2.0 mg/L (36.1%). The final study population comprised men age ≥50 years and women age ≥60 years, with hsCRP levels ≥2 mg/L and LDL-C levels <130 mg/dL, who had no history of cardiovascular disease or stroke.

St. Francis Heart Trial

Approximately 20 000 people were screened for inclusion and exclusion and 5582 people were scanned with computed tomography. A total of 1007 subjects with CAC scores >80th percentile for age and sex were identified and randomly assigned to a study group. The final study population consisted of men and women aged 50 to 70 years, with CAC scores >80th percentile for age and sex, who did not have pre-existing history of cardiovascular disease. Subjects with LDL-C <90 mg/dL were excluded from the trial after initial screening. This was because of the concern that LDL-C <60 mg/dL might alarm primary care physicians, affecting compliance with statin and the study outcome.

Interpretation of Study Populations

The 2 studies had similar patient characteristics: mean age (66 years JUPITER versus 59 years St. Francis Heart Trial), sex distribution (62% men versus 74% men), and Framingham Risk Score (mean 11.6% versus median 11%). By design, the St. Francis Heart trial population had a higher mean LDL-C compared with that of the JUPITER trial (143 versus 104 mg/dL).^23,24 Although the Framingham Risk Score was similar, 77% of the JUPITER trial subjects were non-US populations and Framingham Risk Score may not calibrate as well to non-US populations including Europeans.²⁵ The placebo event rate was significantly higher in the St. Francis Heart trial (9.9% at 4.3 years versus 2.7% at 2 years), suggesting that a high CAC may select a higher cardiovascular risk population as compared with high hsCRP.

It is also noteworthy that the study population is much larger in the JUPITER trial compared with the St. Francis Heart trial (17 802 versus 1005). The authors of the St. Francis trial calculated that with an annual cardiovascular event rate of 3%, 472 patients per group will give an 80% power to identify a hazard ratio of 0.85.⁴ It turned out, however, that the annual major cardiovascular event rate was <1% per year.

Both trials had a relatively arbitrary cutoff value of the biomarker. The JUPITER trial defined people with hsCRP ≥2 mg/L as a high cardiovascular risk group. A study from the Multi-Ethnic Study of Atherosclerosis (MESA) population illustrated that hsCRP >2 mg/L is not an independent predictor of cardiovascular events.²⁶ Numerous studies have suggested a better risk prediction with the hsCRP cutoff at ≥3 mg/L,²⁷ a value also endorsed by the American Heart Association/Centers for Disease Control and Prevention scientific statement.²⁸

The St. Francis Heart trial defined the high CAC score group as subjects with CAC scores >80th percentile for age and sex. This strategy resulted in setting a lower calcium score threshold for younger subjects, particularly women, and therefore a significant number of traditionally low-risk patients were included in the study population.²⁴ As of yet, there is no definite agreement on the CAC cutoff value to identify higher cardiovascular risk populations.

Recent data from the MESA sheds light on the expected CAC profile of JUPITER-eligible patients. It has been projected that 47% of the JUPITER-eligible population would have no CAC, 28% would have CAC 1 to 100, and 25% would have CAC >100.²⁶ Approximately 50% of St. Francis participants had hsCRP ≥2 mg/L.

JUPITER Trial

Eligible subjects underwent a 4-week run-in period before randomization to ensure the capacity of long-term protocol compliance. After the run-in period, participants were randomized to either a rosuvastatin 20 mg daily group or a placebo group. Participants were evaluated at 3 and 6 months initially and then every 6 months for follow-up.

St. Francis Heart Trial

Eligible subjects went through a 2-week wash-in period before randomization. After that, participants were randomized to either an atorvastatin 20 mg daily in addition to aspirin (81 mg/d), vitamin C (1000 mg/d), and vitamin E (1000 U/d) group or placebo with aspirin (81 mg/d) group. Participants were followed up every 6 weeks for 3 months and thereafter at 3-month intervals.

Interpretation of Study Interventions

Rosuvastatin offers the greatest LDL-C lowering and high-density lipoprotein increasing efficacy compared with other statins.²⁹ The Statin Therapies for Elevated Lipid Levels Compared Across Doses to Rosuvastatin (STELLAR) trial showed that 20 mg of rosuvastatin reduced LDL-C by a mean of 52.4%, whereas 20 mg of atorvastatin reduced LDL-C by 42.6%.³⁰ This result is in line with the results of the JUPITER trial and the St. Francis Heart trial, which showed LDL-C reductions by a mean of 50% and 43%, respectively. Therefore, the differential statin potency must be accounted for in the interpretation of the overall study results. Furthermore, the atorvastatin dose in the St. Francis Heart trial was reduced to 10 mg daily when the LDL-C level fell <50 mg/dL. The less aggressive LDL-C reduction with statin may have attenuated event rate reduction in the St. Francis study.³¹

In the St. Francis Heart trial, the atorvastatin treatment arm received high doses of vitamins C (1 g/d) and E (1000 IU/d) along with atorvastatin based on the theoretical benefit of antioxidant treatment.^4,32 However, some studies have demonstrated that coadministration of antioxidant vitamins C and E with statin may attenuate the effect of the statin in cardiovascular outcomes.^33,34 These studies suggest that antioxidant use in the St. Francis Heart trial may have diminished the effects of statin therapy.

It is also noted that both the statin treatment arm and placebo arm in the St. Francis Heart trial received aspirin 81 mg daily. It was deemed unethical not to give aspirin to subjects with elevated CAC scores who may be at increased risk of cardiovascular disease.⁴ Given that the annualized cardiovascular event rate in the placebo arm was lower than expected (2% for the composite end point and <1% for major cardiovascular events),³⁵ it is likely that aspirin therapy might have reduced the power of the St. Francis study.

JUPITER Trial

The JUPITER trial was terminated early after a median follow-up of 1.9 years because of significantly favorable primary outcomes in the rosuvastatin treatment group (1.6% event rate in statin group versus 2.7% in placebo group, 1.1% absolute risk reduction, and 44% relative risk reduction; P<0.00001). The hazard ratio and NNT for rosuvastatin in 2 years were 0.56 and 95, respectively. Extrapolation of the study results outside of the follow-up enabled calculation of an NNT at 5 years of 25 using the method of Altman and Andersen.³⁶

St. Francis Heart Trial

At a mean of 4.3 years of follow-up, treatment reduced total cholesterol by 6.5% to 30.4% (P<0.0001) and LDL-C by 39.1% to 43.4% (P<0.0001). However, the St. Francis Heart trial researchers failed to show a statistically significant difference in the incidence of major cardiovascular events (6.9% versus 9.9%; P=0.08), despite a large absolute risk reduction (3%). Importantly, in a post hoc subgroup analysis, participants with the highest CAC scores (>400) had the greatest benefit from statin treatment (8.7% versus 15.0% event rate, 6.3% absolute risk reduction, 42% relative risk reduction; P=0.046) with associated NNT of 17 at 4 years (14 at 5 years).²⁴

Interpretation of Study Results

It is now firmly established that statins benefit a wide variety of patients.¹ Indeed, both study results are in agreement with a recent meta-analysis of previous randomized control trials that showed a linear relationship between LDL-C reduction and cardiovascular events reduction.¹ (Figure 2) Given the fact that statin therapy has benefits even in people with LDL-C as low as 80 mg/dL,³⁷ the cardiovascular risk reduction shown in both the JUPITER trial and the St. Francis Heart trial (with median LDL-C 108 mg/dL and mean 143 mg/dL, respectively) were predictable without the use of hsCRP and CAC.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Linear relationship between low-density lipoprotein-cholesterol (LDL-C) reduction and cardiovascular events reduction.¹ Cardiovascular risk reductions demonstrated in both trials were in agreement with previous statin trials. 4S indicates Scandinavian Simvastatin Survival Study³⁸; AFCAPS/TexCAPS, Air Force/Texas Coronary Atherosclerosis Prevention Study³⁹; ALERT, Assessment of Lescol in Renal Transplantation⁴⁰; ALLHAT-LLT, Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial⁴¹; ASCOT-LLA, Anglo-Scandinavian Cardiac Outcomes Trial-Lipid Lowering Arm⁴²; CARE, Cholesterol And Recurrent Events⁴³; GISSI, Gruppo Italiano per lo Studio della Sopravvivenza nell’Infarto Miocardico⁴⁴; HPS, Heart Protection Study⁴⁵; LIPS, Lescol Intervention Prevention Study,⁴⁶; PROSPER, PROspective Study of Pravastatin in the Elderly at Risk⁴⁷; and WOSCOPS, West of Scotland Coronary Prevention Study.⁴⁸

The JUPITER trial did not demonstrate effect modification by baseline hsCRP levels. A Food and Drug Administration (FDA)-mandated subgroup analysis of the JUPITER trial revealed that subjects with hsCRP >4.2 mg/L had lower relative risk reduction with statin therapy than those with hsCRP ≤4.2 mg/L (relative risk reduction, 29% versus 58%; P=0.015).^23,49 Moreover, a post hoc analysis of the trial demonstrated that the treatment benefit was confined to people with ≥1 additional risk factor.^23,49 The Anglo-Scandinavian Cardiac Outcome Trial (ASCOT) also showed that while hsCRP is a mild predictor of baseline risk,⁵⁰ there is no improvement in predicting response to statins when adding hsCRP to LDL-C.⁵¹ Thus, none of the JUPITER, ASCOT, or the above-mentioned Heart Protection Study showed evidence of effect modification with elevated hsCRP.

In contrast to the JUPITER trial, the St. Francis Heart trial raises the possibility of effect modification despite being a negative study. In subjects with a baseline CAC score >400, who comprised 47% of the study population in the St. Francis Heart trial, the reduction in incidence of major cardiovascular events was higher than that in subjects with a CAC <400 (8.7% versus 15%; P=0.046; NNT=17 at 4 years).²⁴ However, the possible effect modification with CAC cannot be viewed as definitive in the setting of an overall nonsignificant trial result.

As mentioned, the sample size of the JUPITER trial was much larger than that of the St. Francis Heart trial (17 802 versus 1007). Given that the St. Francis Heart trial had a higher absolute risk reduction (3% versus 1.1%), the sample size is likely to be the main explanation for the nonsignificant P value in the St. Francis heart trial. In exploratory analysis, calculation of the estimated P value in a theoretical St. Francis Heart trial with a similar sample size as JUPITER gives a P value of <0.00001.