Coronary Calcium Better than CRP to Stratify Risk
Download the PDF here
By Todd Neale, Senior Staff Writer, MedPage Today
Published: August 18, 2011
Among asymptomatic individuals with normal LDL cholesterol levels and elevated high-sensitivity C-reactive protein (hsCRP), measuring the burden of calcium in the coronary arteries with cardiac CT appears to stratify the risk of cardiovascular disease, researchers found.
The estimated rate of cardiovascular disease events through a median of 5.8 years was 2.12% for those with a coronary artery calcium (CAC) score of 0, 4.86% for those with a score of 1 to 100, and 13.65% for those with a score greater than 100, Michael Blaha, MD, of Johns Hopkins University in Baltimore, and colleagues reported in the Aug. 20 issue of The Lancet.
Looking further at patients with both low and high levels of hsCRP, the researchers found that CAC score -- but not hsCRP -- was associated with the risk of coronary heart disease events and overall cardiovascular events.
CAC score "could be used to target subgroups of patients who are expected to derive the most, and the least, absolute benefit from statin treatment," Blaha and colleagues wrote. "Focusing of treatment on the subset of individuals with low LDL cholesterol with measurable atherosclerosis might represent a more appropriate allocation of resources, reduce overall healthcare cost, and prevent the occurrence of a similar number of events."
But evidence from randomized trials that such an approach -- which remains controversial -- would actually reduce cardiovascular events in asymptomatic individuals is lacking.
Steven Nissen, MD, of the Cleveland Clinic, pointed out in an email to ABC News and MedPage Today that calcium scanning has never been shown to reduce the risk of MI or death. He called the practice "one of the worst examples of 'medicine gone wild,'" pointing to unacceptable radiation doses and the greater use of unnecessary catheterization and stenting following a scan.
Other physicians contacted, however, supported the use of CAC scoring.
"Use of CAC is helpful in saying if the process of atherosclerosis has started, and is helpful in choosing between lifestyle intervention versus lifestyle intervention plus statin therapy," wrote Christopher Cannon, MD, of Brigham and Women's Hospital in Boston.
The basis of the current analysis was the JUPITER trial, which showed that treatment with rosuvastatin (Crestor) reduced MI, stroke, and cardiovascular death in patients with normal LDL cholesterol levels but elevated hsCRP.
Blaha and colleagues identified a JUPITER-like cohort in the Multi-Ethnic Study of Atherosclerosis (MESA) to determine whether measuring CAC would stratify patients according to cardiovascular risk and potentially identify a subgroup of patients who would most benefit from statin therapy.
The analysis included 950 MESA participants who had normal LDL cholesterol levels (less than 130 mg/dL) and elevated hsCRP (2 mg/L or more). All underwent two noncontrast CT scans to measure the calcium.
The authors noted that the average measured dose of radiation in a CAC scan was 0.89 mSv in MESA, while the average dose overall with modern technology ranges from 0.5 to 1.5 mSv.
Nearly half of the patients (47%) did not have any coronary calcium (a score of 0). Another 28% had scores of 1 to 100 and 25% had scores greater than 100.
Per 1,000 person-years, rates of coronary heart disease events increased from 0.8 for patients with a CAC score of 0 to 20.2 for patients with a score greater than 100. The patterns were similar for all cardiovascular events.
Through a median follow-up of 5.8 years, 74% of all coronary heart disease events and 60% of all cardiovascular events occurred in the patients with a CAC score greater than 100.
Applying the risk reduction accompanying rosuvastatin treatment observed in JUPITER (a 44% relative reduction in the risk of MI, stroke, and cardiovascular death), Blaha and colleagues calculated that the five-year number needed to treat (NNT) to prevent one coronary heart disease event was 549 for patients with a CAC score of 0, 94 for those with intermediate CAC scores, and 24 for those with scores greater than 1oo.
The NNTs for all cardiovascular events were 124, 54, and 19, respectively.
"These results have important implications for future guidelines and public health discussions aimed at improving the efficiency of statin use in primary prevention," according to the authors, who noted that studies have suggested that asymptomatic patients with a CAC score of 0 can be treated less aggressively with an emphasis on low-cost lifestyle interventions.
Blaha and colleagues acknowledged that a cost-benefit analysis is needed in patients with both low and high hsCRP to determine whether using CAC scores to guide statin treatment improves outcomes. They also acknowledged that CAC scoring has both advantages and disadvantages compared with hsCRP.
Advantages of CAC scoring include the fact that it is a direct measure of the burden of atherosclerosis, it has small variability on repeated testing, and it has consistent thresholds of risk in different populations. Thresholds for hsCRP vary by sex and ethnic origin.
Disadvantages include radiation exposure, a risk of incidental findings leading to further imaging, and a higher cost compared with hsCRP.
Still, CAC scanning can be performed for less than $100 in many centers, the researchers noted.
"Now that CAC scoring is so inexpensive, one can make the case of testing this in most patients at intermediate risk," commented Carl Lavie, MD, of the Ochsner Heart and Vascular Institute in New Orleans, in an email.
One obstacle to the wider adoption of CAC scoring for risk stratification "is that use of preventive measures such as statins, blood pressure control, and weight loss do not reduce the coronary calcium, and this is unsettling for many physicians and patients," said Howard Weintraub, MD, clinical director of the NYU Center for the Prevention of Cardiovascular Disease, in an email.
Some physicians want to see the value of the test proven in a randomized trial as well, although that has not stopped all physicians from incorporating it into their practices.
"Although definitive proof of treatment effects is scarce, CAC identifies high cardiovascular risk, and statin therapy is most effective in high-risk patients," Axel Schmermund, MD, and Thomas Voigtlander, MD, of Cardioangiologisches Centrum Bethanien in Frankfurt, wrote in an accompanying editorial.
"In our practice, we therefore focus on CAC ... for expanded risk stratification in asymptomatic patients."
Predictive ability of coronary artery calcium and CRP - Editorial
The Lancet August 20, 2011
Axel Schmermund, Thomas Voigtlander
Cardioangiologisches Centrum Bethanien, CCB, Im Prufl ing 23,
D-60389 Frankfurt am Main, Germany
In the general population, less than 10% of healthy adults aged 25-74 years have no modifiable cardiovascular risk factors;1 therefore, risk of cardiovascular disease can potentially be improved in most people. Statin therapy for lowering cholesterol is an important cornerstone of risk reduction. The absolute benefit of statin treatment increases with increasing patient risk; thus, risk stratification of asymptomatic patients is mandatory in clinical practice.
Although accurate identification of future cardiovascular disease risk is difficult when overall risk is low, the Framingham risk score and other global risk scores offer a meaningful approximation.2 Such algorithms now allow for a practical approach towards risk stratification, translating statistical data into quantification of an individual's global risk. However, many uncertainties remain: because more than 40% of individuals have an intermediate risk of 10-20% in 10 years, treatment options are restricted; the scores are best at predicting long-term risk even though substantial risk factor changes can occur over time; and levels of absolute risk differ across cultural and ethnic groups. Thus, individual risk stratification needs further improvement in asymptomatic adults.
C-reactive protein (CRP) and coronary artery calcium (CAC) are among the most thoroughly examined measures available for expanded risk stratification. CRP is an acute-phase reactant synthesised mainly in the liver. From an evolutionary perspective, the teleological function of CRP might have been as part of the innate immune system, promoting complement activation and antigen presentation.3 Within the range of normal values, easily available and highly sensitive assays detect even small amounts of CRP, thus rendering it an attractive and sensitive biomarker of subclinical inflammation. Because atherosclerosis is an inflammatory disease, CRP has been associated with imminent activation of the disease and increased patient vulnerability (ie, the patient is at increased risk of a cardiovascular event). Therefore, a logical option was to investigate the practical applicability of this biomarker. The JUPITER trial4 examined the effects of statin therapy in patients with no clinical cardiovascular disease, and with LDL in the normal range, but higher than average concentrations of CRP. Reduction of clinical events was of such a magnitude in this group (44% reduction in relative risk) that the trial was ended after only 1·9 years instead of 5 years as first planned. JUPITER did not include a control group of patients with low CRP. Was the beneficial effect of the statin therapy in JUPITER due to optimum patient selection by use of CRP?
In The Lancet, the well-designed substudy of the MESA trial by Michael Blaha and colleagues5 presents data that indicate a different conclusion. MESA recruited 6814 unselected participants free of known cardiovascular disease from six centres throughout the USA. The investigators' main objective was to analyse the predictive ability of CAC in asymptomatic patients. 2083 MESA participants were identified who met JUPITER inclusion criteria and whose high-sensitivity CRP (hsCRP) values were 2 mg/L or more. After full adjustment for Framingham risk scores, high CRP did not seem to affect future coronary heart disease events (hazard ratio 0·90, 95% CI 0·54-1·50), whereas the presence of CAC, and particularly CAC scores of more than 100, were strongly predictive of both coronary heart disease (9·35, 4·15-21·1) and overall cardiovascular disease events (4·41, 2·42-8·04).5 Blaha and colleagues used the event reductions noted in JUPITER to estimate that treatment with a potent statin in patients with raised CAC would be highly efficient (numbers needed to treat [NNT] to prevent an event of coronary heart disease: 5-year NNT when any CAC was present was 42, and for CAC scores >100 5-year NNT was 24), whereas treatment of patients with CAC scores of 0 would be unfavourable (5-year NNT was 549). How can this be explained and what are the practical implications?
CAC is a specific expression of coronary atherosclerotic plaque disease, with a linear relation between the extent of CAC and the overall extent of coronary atherosclerosis.6 Although exceptions in patients aged less than 50 years are possible, to find someone with extensive coronary atherosclerosis who has no CAC is highly unlikely.7 Despite the difficulties that have been unresolved for prospective classification of plaques as stable or unstable, CAC does not allow such distinctions.8 Clinical studies8-10 support the conclusion that increased CAC signifies increased amounts of plaque, and thus an increased risk.
Conversely, with no CAC, plaque is absent or scarce, and risk is low.
Neither CAC nor CRP have a significant causal role in cardiovascular events.8, 11 Unlike CAC, the association of CRP with cardiovascular events seems to vary between different patient subgroups defined by level of risk and ethnicity.12 Various inflammatory and other stimuli and components of the metabolic syndrome might increase CRP. Accordingly, in some patients with low CAC scores, CRP could be a general marker of poor health.13 However, CAC is a much more specific expression of atherosclerosis, the immediate precursor of cardiovascular events, which probably explains the better predictive ability of CAC than of CRP for both coronary heart disease and cardiovascular disease.5, 13
We cannot be content with statistical significance, and a practical approach towards expanded risk stratification should be established. This scenario has been achieved for high CRP in the JUPITER trial for the benefits of statin therapy. Except for one preliminary trial,14 such data are unavailable for CAC. Nevertheless, practical application should be reserved for a measure whose statistics are clearly predictive, and as Blaha and colleagues show, this case is much stronger for CAC than for CRP. Although definitive proof of treatment effects is scarce, CAC identifies high cardiovascular risk, and statin therapy is most effective in high-risk patients. In our practice, we therefore focus on CAC and use the algorithm shown in the figure for expanded risk stratification in asymptomatic patients.
The Lancet Pages 684 - 692, 20 August 2011
Associations between C-reactive protein, coronary artery calcium, and cardiovascular events: implications for the JUPITER population from MESA, a population-based cohort study
The JUPITER trial showed that some patients with LDL-cholesterol concentrations less than 3·37 mmol/L (<130 mg/dL) and high-sensitivity C-reactive protein (hsCRP) concentrations of 2 mg/L or more benefit from treatment with rosuvastatin, although absolute rates of cardiovascular events were low. In a population eligible for JUPITER, we established whether coronary artery calcium (CAC) might further stratify risk; additionally we compared hsCRP with CAC for risk prediction across the range of low and high hsCRP values.
950 participants from the Multi-Ethnic Study of Atheroslcerosis (MESA) met all criteria for JUPITER entry. We compared coronary heart disease and cardiovascular disease event rates and multivariable-adjusted hazard ratios after stratifying by burden of CAC (scores of 0, 1-100, or >100). We calculated 5-year number needed to treat (NNT) by applying the benefit recorded in JUPITER to the event rates within each CAC strata.
Median follow-up was 5·8 years (IQR 5·7-5·9). 444 (47%) patients in the MESA JUPITER population had CAC scores of 0 and, in this group, rates of coronary heart disease events were 0·8 per 1000 person-years. 74% of all coronary events were in the 239 (25%) of participants with CAC scores of more than 100 (20·2 per 1000 person-years). For coronary heart disease, the predicted 5-year NNT was 549 for CAC score 0, 94 for scores 1-100, and 24 for scores greater than 100. For cardiovascular disease, the NNT was 124, 54, and 19. In the total study population, presence of CAC was associated with a hazard ratio of 4·29 (95% CI 1·99-9·25) for coronary heart disease, and of 2·57 (1·48-4·48) for cardiovascular disease. hsCRP was not associated with either disease after multivariable adjustment.
CAC seems to further stratify risk in patients eligible for JUPITER, and could be used to target subgroups of patients who are expected to derive the most, and the least, absolute benefit from statin treatment. Focusing of treatment on the subset of individuals with measurable atherosclerosis could allow for more appropriate allocation of resources.
Findings from landmark clinical trials1-3 have led to progressive liberalisation of statin use for primary prevention of cardiovascular disease. The JUPITER trial4 led to further liberalisation by showing that some patients with normal concentrations of LDL cholesterol (ie, <3·37 mmol/L) and high-sensitivity C-reactive protein (hsCRP) (≥2 mg/L) benefit from treatment with rosuvastatin. Unfortunately, because modern statin trials enrol low-risk populations, even large reductions in relative risk result in only small reductions in absolute risk. Thus, many patients who are newly eligible for statins will not accrue a net benefit from treatment. Personalised assessment of cardiovascular risk is still needed.
Coronary artery calcium (CAC) detected by cardiac CT estimates the burden of coronary atherosclerosis and is effective for further stratification of risk in patients who are asymptomatic.5 The absence of CAC in an asymptomatic adult nearly excludes clinically important coronary atherosclerosis, and is associated with a mortality rate of about 1% in 10 years.6, 7 By contrast, substantially increased CAC is associated with a rise of almost ten times in risk of adverse coronary events after multivariable adjustment.8 Furthermore, CAC can improve the classification of patients into appropriate risk groups for clinical decision making.9
We sought to establish whether tests for CAC could identify a subgroup of patients eligible for JUPITER who would be expected to derive the most or the least benefit from statin treatment. In view of estimates based on findings from JUPITER that 6·5 million individuals in the USA would be newly eligible for statins,10 these results have important implications for guidelines and public health discussions aimed at improving the efficiency and cost-effectiveness of statin use in primary prevention.
Furthermore, we aimed to directly compare CAC with hsCRP as additional markers to identify risk in individuals eligible for JUPITER, independent of hsCRP inclusion criteria. Such comparative-effectiveness analyses examining the incremental predictive value of tests in their intended target populations are crucial for their appropriate use.
Median age of the total study population (N=2083) was 67 years (IQR 61-73). Overall, 835 (40%) were women, with mean calculated 10-year Framingham risk of 9·7% (SD 7). Median hsCRP of the total study population was 1·8 mg/L (IQR 0·78-4·0). 1133 (54%) participants had hsCRP less than 2 mg/L, and 950 (46%) had hsCRP 2 mg/L or more (MESA JUPITER population). Individuals in the MESA JUPITER subgroup were more likely to be women and either African-American or Hispanic, with more features of the metabolic syndrome than those with hsCRP less than 2 mg/L (table 1).
The MESA JUPITER population closely resembled the placebo group in the JUPITER trial (webappendix p 2). Median age of patients in the JUPITER placebo group was 66 years (IQR 60-71), mean calculated 10-year Framingham risk of 10%, and median hsCRP was 4·3 mg/L (IQR 3·0-7·8). The MESA JUPITER population had more women than the JUPITER population (51% vs 38%) because of its population-based recruitment with similar initial enrolment by gender, coupled with the higher concentrations of hsCRP in women.
444 (47%) patients in the MESA JUPITER population had a CAC score of 0. Of those with CAC, 267 (28%) had scores 1-100, and 239 (25%) had scores more than 100. The number needed to scan to identify one individual with a CAC score of 0 was 2, and to identify one individual with a CAC score more than 100 was 4. The frequency of increased CAC burden was similar in the low hsCRP group (p=0·09, webappendix pp 3-6). Prevalence of CAC differed according to sex. 259 (53%) women had a CAC score of 0 compared with 185 (40%) men, and 97 (20%) of women had a score of more than 100 compared with 142 (31%) men.
Table 2 shows the frequency of cardiovascular disease events and coronary heart disease events, the corresponding event rates per 1000 person-years, and the multivariable-adjusted HRs associated with prevalence and burden of CAC in MESA JUPITER participants. Event rates for coronary heart disease and cardiovascular disease were low when CAC scores were 0 and high when CAC scores were more than 100 (table 2). Only 6% of all coronary heart disease events and 17% of all cardiovascular disease events were in the individuals with scores of 0 (table 2). Almost 75% of all coronary heart disease events, and about 60% of all events of cardiovascular disease, were in the 25% of participants with scores more than 100 (table 2).
The presence of CAC was associated with an HR of 11·0 (95% CI 2·51-48·5) for coronary heart disease, and 3·20 (1·41-7·24) for cardiovascular disease in the MESA JUPITER population in the fully adjusted model. We noted a graded increase in events for both diseases with increasing burden of CAC (table 2).
Figure 2 shows Kaplan-Meier estimates of event-free survival for coronary heart disease and cardiovascular disease for the MESA JUPITER population by CAC burden. Table 3 shows the Kaplan-Meier failure (event) function. From these estimates, the 5-year NNT to prevent an event of coronary heart disease was 549 for CAC score 0, 94 for scores 1-100, and 24 for scores more than 100. The corresponding 5-year NNT to prevent an event of cardiovascular disease was 124, 54, and 19, respectively (table 3). Webappendix p 1 shows the results of the sensitivity analysis.
In the total study population, overall event rates were similar in the low (<2 mg/L) and high (≥2 mg/L) hsCRP groups for coronary heart disease (7·6 vs 6·4 per 1000 person-years, p=0·47) as were event rates for cardiovascular disease (10·1 vs 10·4 per 1000 person-years, p=0·87). Figure 3 shows Kaplan-Meier plots stratified by hsCRP status. hsCRP status did not predict coronary heart disease events (HR 0·98, 95% CI 0·62-1·57) or cardiovascular disease events (1·15, 0·78-1·68) after adjustments for age, sex, and race. By contrast, presence of CAC was a strong predictor of both coronary heart disease (6·65, 2·99-14·78) and cardiovascular disease (3·06, 1·82-5·13) in similarly adjusted models. CAC prevalence and increased CAC burden were significant predictors of events after full multivariable adjustment (table 4).
Increased CAC burden led to similar increases in absolute coronary heart disease and cardiovascular disease events in both the low and high hsCRP groups (figure 4). We recorded no evidence of interaction between hsCRP status and CAC burden (p=0·71), or of residual confounding with hsCRP with use of dichotomised hsCRP status (low and high). Median hsCRP in the MESA JUPITER population was 4·54 mg/L (IQR 2·77-10·6) with cardiovascular disease events and 4·25 mg/L without (2·96-7·71, p=0·61). Median hsCRP in the total study population was 1·73 mg/L (IQR 0·84- 4·15) in participants with cardiovascular disease events, and 1·78 mg/L (0·78-3·98) in those without events (p=0·67). 48 (68%) of the 71 coronary heart disease events were classed as so-called hard coronary heart disease events (myocardial infarction, resuscitated cardiac arrest, or death from coronary heart disease), and 79 (67%) of 118 cardiovascular disease events (hard coronary heart disease events plus stroke [not transient ischaemic attack] or stroke death). No differences were recorded in the predictive value of CAC or hsCRP when hard events were substituted for all coronary heart disease or cardiovascular disease events (data not shown).
As statin use is extended to low-risk populations, accurate assessment of absolute risk becomes crucial to measure the net value of treatment. Our findings show that nearly half of the MESA JUPITER population had no CAC, had a very low event rate, and an unfavourable estimated 5-year NNT of 549 to prevent one coronary heart disease event. By contrast, most coronary heart disease events (74%) were in the small (25%) group of MESA JUPITER patients with CAC scores greater than 100. With these scores, the estimated 5-year NNT was small at 24 for coronary heart disease and 19 for cardiovascular disease. These results have important implications for future guidelines and public health discussions aimed at improving the efficiency of statin use in primary prevention (panel).
Research in context
Searches of PubMed and Google Scholar for articles published from 2000 to 2011, supplemented by hand searches of reference lists of review articles and meta-anlyses, yielded high-quality summaries of the independent predictive values of coronary artery calcium (CAC) and high-sensitivity C-reactive protein (hsCRP). Methodology of these studies differs sufficiently from the present report to preclude meta-analysis.
The highest quality articles suggest that both CAC and hsCRP improve prediction of cardiovascular disease risk beyond global risk assessment algorithms.8, 14 Although no systematic review compares CAC with hsCRP for risk prediction, smaller studies have suggested that CAC is a stronger predictor than hsCRP.15, 16 Our study from the Multi-Ethnic Study of Atherosclerosis, which includes measurements of baseline CAC and hsCRP, and 6-year follow-up, confirms the excellent prognosis associated with CAC scores of 0, and extends this finding to the population eligible for JUPITER. Our conclusion that CAC is a stronger predictor of cardiovascular events than hsCRP is consistent with previous reports,15, 16 and we extend these findings to the population with low LDL cholesterol (<3·37 mmol/L). Our results are consistent with the hypothesis that focus of treatment on the subset of individuals who have low LDL cholesterol with measurable atherosclerosis could represent a more appropriate allocation of resources, and reduce overall health-care cost, while preventing a similar number of events.
Current guidelines for primary prevention lend support to the use of statins to treat increased cholesterol in individuals deemed high risk by traditional risk scoring. Future guidelines might incorporate the recommendation for statin treatment in patients with normal cholesterol who are at increased risk because of another risk factor or biomarker (such as hsCRP). In view of our results, CAC should be strongly considered in these patients; this supports the IIA recommendation for CAC screening in the updated American Heart Association guidelines for testing in adults who are asymptomatic.17
Because of the inflammatory hypothesis of atherothrombosis, increased hsCRP might provide a mechanistic link to individuals who will receive the greatest benefit from statins.18 Without a biomarker control group of individuals with hsCRP concentrations less than 2 mg/L in JUPITER, whether such low hsCRP patients would have similarly benefited is impossible to determine. Secondary analyses from JUPITER have shown that the reduction in relative risk with rosuvastatin was remarkably consistent, and not graded, across increased concentrations of hsCRP.19 Secondary analysis of the Heart Protection Study showed that statins achieve a similar relative risk reduction at all concentrations of hsCRP, including in patients with low hsCRP.20 Therefore, the benefit of hsCRP testing seems to rely solely on its generally consistent association with slightly increased absolute risk, and thus anticipated high absolute benefit from treatment.19
We noted that the presence of CAC identifies both absolute and relative risk of coronary heart disease over a much wider range than an hsCRP of greater than or equal to 2 mg/L. Although CAC predicts cardiovascular disease-including stroke-less strongly than it does coronary heart disease, it is still better than use of hsCRP (2·57 vs 1·08 in fully adjusted models). Our finding that hsCRP does not effectively identify risk has been noted in other studies,21 but is in contrast to the moderate independent predictive value of hsCRP in the largest meta-analysis (relative risk 1·63, adjusted for sex and age).14 Reasons for the failure of hsCRP to predict risk in MESA might include the various ethnic origins of patients in the cohort and the use of the fixed JUPITER cutoff of 2 mg/L, which does not account for the highly different distributions of hsCRP across sex and race in a highly diverse population.
CAC has both advantages and disadvantages compared with hsCRP. CAC is a direct measure of the burden of atherosclerosis-the precursor lesion for most coronary heart disease events-and is best regarded a measure of disease rather than a risk factor. Indeed, the progression of CAC is a strong predictor of mortality.22 Another advantage is the small variability when the measurement is repeated.23 Additionally, CAC has consistent thresholds of risk in different populations,8 whereas hsCRP varies greatly by sex and ethnic origin, with few data for variations in risk thresholds.24, 25 Although a disadvantage of CAC is radiation exposure (the average measured dose of radiation was 0·89 mSv in MESA), the dose with modern technology is low (0·5-1·5 mSv compared with background radiation of 3 mSv per year). Furthermore, incidental non-cardiac findings, such as lung nodules more than 4-6 mm in diameter, generally lead to referral for imaging follow-up at 6-12 months, despite no proven mortality benefit for this follow-up. CAC is more expensive than hsCRP testing; however, many metropolitan areas in the USA charge less than US$100 for CAC testing. Although hsCRP has possible value in monitoring the potency of the effect of statin treatment,26 no data or biologically plausible mechanisms are available to suggest that statins reduce CAC.27
Published work15, 16 has suggested that a combination of hsCRP and CAC might be better than use of either method alone in select patients. Park and colleagues15 followed up 967 individuals without diabetes for a mean of 6·4 years, and showed that most risk resided with CAC, with very high hsCRP (>4·05 mg/L) providing slight improvement in incremental risk. Similarly, data from the Heinz-Nixdorf Recall study16-a large cohort study with a design similar to MESA-showed that improvement in prediction and discrimination of coronary risk was driven mostly by CAC, with hsCRP more than 3 mg/L (the JUPITER cutoff point of 2 mg/L was not studied) providing mild incremental improvement mainly versus hsCRP less than 1 mg/L in people with very low CAC scores. Importantly, CAC and hsCRP could identify distinct mechanisms of risk. CAC, but not hsCRP,28 identifies overall burden of coronary atherosclerosis, although emerging data indicate that hsCRP might provide some insight into the stability of the coronary plaque.29
Our results have important public health implications. MESA patients with no coronary calcification who are eligible for JUPITER had a very low rate of coronary heart disease events of less than 1 per 1000 person-years, corresponding to about a 1% 10-year event rate, consistent with data showing excellent prognosis when CAC scores are 0.6, 7 Reports have suggested that asymptomatic patients with such scores can be treated to less aggressive targets, with less aggressive pharamacotherapy, emphasising low-cost lifestyle interventions.30 The 5-year NNT to prevent one coronary heart disease event of 549 in this study when CAC is 0 seems to support a conservative strategy. Indeed, this NNT exceeds the 4-year NNT of 255 for new-onset diabetes recorded with statin use in a meta-analysis.31
Similar to many studies,6, 8 most events in MESA JUPITER participants were in those few with CAC scores greater than 100. The rate of 20-26 events per 1000 person-years in this group puts them within the conventional high-risk designation of more than a 20% 10-year risk. On the basis of these findings, restriction of therapy to those with scores of more than 100 (about a quarter of the JUPITER population) would result in treatment of a subgroup in whom nearly 75% of all coronary heart disease events would occur. If statin therapy were limited to those with CAC (about half the JUPITER population), treatment of a subgroup who have 95% of coronary heart disease events would occur over 6 years. Event rates in MESA (table 2) were lower than those in the JUPITER placebo group (13·6 events of cardiovascular disease per 1000 patient-years) and the ARIC JUPITER population (15·7).32 Despite this finding, the 5-year NNT of 19 for cardiovascular disease in the MESA JUPITER population with CAC greater than 100 is lower than the overall estimate in JUPITER (5-year NNT 25, extrapolated from median follow-up of 1·9 years)33 and in ARIC (38, adjusted from mean follow-up of 6·9 years).32
In the short term, a cost-benefit analysis is needed to explore the potential effect of allocation of statins guided by CAC in populations with both high (JUPITER eligible) and low hsCRP. Findings from a similar study for hsCRP34 showed that hsCRP screening was not more cost effective than traditional risk-based allocation of statins. The EISNER study35 suggested a potential cost saving with CAC screening, with substantially reduced downstream spending in the large group of participants with CAC scores of 0.
Many believe that a clinical trial is needed before CAC can be widely endorsed to stratify risk in adults for whom treatment decisions are unclear.36 Such a trial could be approached in several ways. One design would aim to show overall cost savings with non-inferior clinical outcomes (increased treatment efficiency) when CAC scoring is used to allocate statin treatment. Another design would randomly assign patients to CAC screening versus traditional risk assessment, thus aiming to show a net treatment benefit when those with elevated CAC receive an additional multifaceted, dosed-intensity lifestyle and pharmacotherapy intervention. However, there are challenges to such a trial design, including cost and insufficient knowledge about key assumptions (eg, whether CAC testing would improve adherence to therapies).37 Another potential design, analogous to the JUPITER study design, would be to randomly assign patients with increased CAC, but with Framingham 10-year risk estimates of coronary heart disease of less than 10% to treatment or no treatment. However, this design could be regarded as unethical in view of the strong relations between raised CAC and future cardiovascular events.
The main limitation of this analysis is the uncertainty in application of the reduction in relative risk noted in JUPITER to a separate population for the estimation of NNT. For example, whether patients with increased CAC obtain an equivalent benefit with statins compared with those with low or no CAC is unknown. The only available data are from a post-hoc analysis of the St Francis Heart Study,38 which showed that atorvastatin 20 mg significantly lowered events in patients with CAC score more than 400, with non-significant lowering of events in those with lower scores. As such, our NNT results should be regarded as hypothesis generating. How the greater prevalence of women than men in our population affects the overall results is unclear.
In conclusion, CAC seems to further stratify risk in patients who meet eligibility criteria for JUPITER, and might be used to target a subgroup of patients expected to derive the most and the least absolute benefit from treatment. Focusing of treatment on the subset of individuals with low LDL cholesterol with measurable atherosclerosis might represent a more appropriate allocation of resources, reduce overall health-care cost, and prevent the occurrence of a similar number of events.