This meta-analysis was reported in accordance with the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) Statement; the PRISMA checklist is available in the supplementary data.12 The original study protocol was prospectively submitted for registration in PROSPERO and protocol amendments have been updated (registration code CRD42020171930).

All published clinical studies including patients without a previous ASCVD who underwent CAC score assessment were evaluated for inclusion in this meta-analysis. We considered for inclusion randomized clinical trials (RCT) or observational studies reporting ASCVD outcomes (defined as a composite endpoint including at least myocardial infarction or a proxy for myocardial infarction such as coronary revascularization) and lipid-lowering therapy status stratified by CAC values. Studies reporting the main study outcomes described in insufficient detail or not written in the English language were excluded. The main study outcome was ASCVD occurrence at last follow-up. ASCVD definition for each included study is reported in table 1. Patients were categorized by CAC strata (CAC 0, CAC 1-100, CAC> 100) and by lipid-lowering therapy status (yes vs no). The impact of lipid-lowering therapies on ASCVD occurrence stratified by CAC categories was evaluated. The ASCVD risk stratification ability of CAC score, overall and stratified by lipid-lowering therapy status was also evaluated.

Five authors (E. Elia, F. Bruno, F. Angelini, G. Gallone, and P.P. Bocchino) independently searched EMBASE, MEDLINE/PubMed, and the Cochrane Central Register of Controlled Trials (CENTRAL) using a combination of the following free-text words: “calcium artery score”, “calcium score”, “CAC”, “statin”, “lipid lowering”, “preventive therapy” (detailed search strategy in the supplementary data) from inception to June 15, 2021. Backward snowballing was also performed (no additional studies found).

All authors independently assessed identified studies for possible inclusion. Nonrelevant articles were excluded based on the title and abstract. Two investigators (U. Annone, and F. Piroli) independently extracted data on study designs, measurements, patient characteristics, and outcomes using a standardized data extraction form. Conflicts regarding inclusion and data extraction were discussed and resolved with another investigator (L. Franchin). Data collection included authors, year of publication, inclusion and exclusion criteria, sample size, baseline clinical features of patients, observed adverse events, and medical treatment, as available. To improve data extraction, supplementary data and pertinent substudies were also examined.

Two independent reviewers (M. Bertaina, and E. Elia) assessed the risk of bias (low, intermediate, or high) of the included studies following the Agency for Healthcare Research and Quality recommendations.16

The analysis was by aggregate data. Cumulative event rates for study endpoints were obtained and reported. Pooled effect estimates of the outcomes were calculated as the weighted mean difference using a random effects model and are presented with 95% confidence intervals (95%CI). Subgroup analysis was performed including only RCTs and studies with multivariate adjustment. Heterogeneity across studies was assessed using Cochrane Q statistics and I2 values. I2 values of less than 25% indicate low heterogeneity, 25% to 50% moderate heterogeneity, and greater than 50% high heterogeneity. Statistical significance was set at P <.05 (2-sided). Statistical analyses were conducted with RevMan 5.3 (The Nordic Cochrane Centre, The Cochrane Collaboration, 2014).

Forest plots for the risk of ASCVD occurrence with vs without lipid-lowering therapy in the overall population and stratified by CAC categories are reported in figure 2. In the overall primary prevention population, numerically reduced ASCVD events were observed among patients on lipid-lowering therapy (hazard ratio, 0.84; 95%CI, 0.68-1.04, I2=54%), a difference that became statistically significant when only adjusted data were pooled (hazard ratio, 0.59; 95%CI, 0.38-0.91, I2=85%), figure 3). A significant interaction was observed among CAC subgroups (P=.004, figure 2), so that lipid-lowering therapy was associated with reduced occurrence of ASCVD in patients with CAC> 100 (odds ratio [OR], 0.69; 95%CI, 0.53-0.91, I2=48%), but not in patients with CAC 1-100 or CAC 0. Results were consistent when adjusted data-only were pooled (figure 3). An assessment of the plausibility of the observed subgroup differences18 is detailed in the supplementary data.

Figure 2.

Summary forest plots for the observed risk of atherosclerotic cardiovascular disease occurrence with vs without lipid-lowering therapy stratified by CAC categories. 95%CI, 95% confidence interval; CAC, coronary artery calcium; df, degrees of freedom; F, females; M, males.

(0.6MB).

Figure 3.

Summary forest plots for the adjusted risk of atherosclerotic cardiovascular disease occurrence with vs without lipid-lowering therapy stratified by CAC categories. 95%CI, 95% confidence interval; CAC, coronary artery calcium; df, degrees of freedom; SE, standard error.

(0.32MB).

ASCVD incidence rates for each study are reported in table 2 of the supplementary data. A graded increase in ASCVD occurrence was observed for increasing CAC strata (table 3). Compared with patients with CAC 1-100, patients with CAC 0 were at lower (OR, 0.56; 95%CI, 0.44-0.67), and patients with CAC> 100 were at higher (OR, 2.45; 95%CI, 2.15-2.75) risk of ASCVD. The results were consistent both in patients with and without lipid-lowering therapy (table 3) and remained similar in a sensitivity analysis limited to patients on lipid-lowering therapy prior to CAC assessment (table 3 of the supplementary data).

Among individuals with no detectable CAC, representing 41% to 57% of individuals eligible for lipid-lowering therapy, the 10-year actual ASCVD event rate was much lower than predicted, ranging between 1.5% and 4.9%.20 Similarly, among individuals with ≥ 3 risk factors, 35% had no detectable CAC and a 7-year ASCVD event rate of around 3/1000 person-years.21

Our results, not accounting for the clinical risk profile, consistently suggest that lipid-lowering in this population therapy may not be beneficial. Caution is warranted in translating this finding to specific subsets, including smokers, individuals with severe familial hypercholesterolemia, those with a strong family history of ASCVD and those with a 10-year ASCVD estimated risk ≥ 20%, who demonstrated substantial 10-year actual ASCVD risk despite no detectable CAC.11,20,22,23

Regarding young individuals (<45 years), no detectable CAC is highly prevalent (consistently more than 90% across the literature); accordingly its role as a screening strategy in this subset has been questioned.7 When available, a CAC 0 entails a very benign prognosis with an estimated 10-year mortality of 0.4%.21 Nevertheless, considering the very long-term expected lifespan of this population and that ASCVD event risk depends on cumulative prior exposure to low-density lipoprotein cholesterol,24 it remains to be established whether early initiation of lipid-lowering therapy may translate into a very long-term clinical benefit in young hypercholesterolemic individuals.

We found no significant treatment benefit among patients with CAC 1-100, including in the analysis adjusted for clinical risk factors (aOR, 0.64; 95%CI, 0.36-1.13; P=.12, I2=74%). However, the numerical trend toward a benefit of lipid-lowering therapy (against a background of previous studies showing that 10% ASCVD actual risk in patients with CAC 1-100 varies widely between 3.8% and 14.3% according to sex, age and ethnicity9) suggests that, in this CAC range, ASCVD clinical risk estimation is warranted to indicate lipid-lowering therapy. This concept has been empirically embraced by the American College of Cardiology/American Heart Association guidelines, which favor lipid-lowering therapy initiation only in adults> 55 years of age, when CAC scores of 1 to 99 are found.23

A CAC score> 100 identifies individuals at the higher end of the cardiovascular risk spectrum despite a low burden of traditional risk factors. Specifically, it translates into an 10-year actual risk of ASCVD> 7.5%, regardless of clinically estimated 10-year ASCVD risk.9 Young individuals (< 45 years) with elevated CAC burden had a much higher mortality risk than elderly individuals (> 75 years) with a CAC score of zero.25 Similarly, among individuals with no risk factors, 12% had CAC> 100 and experienced an ASCVD rate of 9.2 per 1000 person-years.21

Our study findings extend these observations further by showing a substantial benefit of lipid-lowering therapy in patients with CAC> 100. A Report and Systematic Review for the US Preventive Services Task Force assessing the benefits and harms of CAC score suggested that the score may inappropriately reclassify individuals not having ASCVD into higher-risk categories, thus prompting unneeded treatment.26 Our analysis does not concur with this concept, since a consistent treatment benefit was observed among these patients and in the analysis adjusted by clinical risk factors.

Of note, the utility of a CAC-guided over clinically-guided treatment strategy in primary prevention seems to apply also to aspirin, for which recent studies adopting meta-analysis data on ASCVD relative risk reduction and bleeding risk seem to suggest that, while aspirin allocation guided by the pooled cohort equations may translate in net harm across all ASCVD risk classes, a strategy complemented by CAC evaluation may identify subsets of patients with a risk trade-off favoring aspirin treatment (ie, patients with CAC> 100 in the setting of low bleeding risk and more than low ASCVD risk).27,28

Finally, CAC-guided compared with clinical risk-guided lipid-lowering therapy appears to have a cost-effective profile.29 The single RCT comparing a CAC-based with a risk factor-based strategy consistently showed improved cardiovascular risk factor control without increased downstream resource use by more appropriate resource allocation to at risk patients.30

To conclude, we observed that the CAC stratification ability was preserved among patients on lipid-lowering therapies, both among those already on treatment at CAC assessment and among those starting treatment thereafter.

Some authors have raised concerns that the plaque-stabilizing effect of statins, which is reflected by an increase in CAC score, might affect the risk stratification ability of CAC score assessment among patients on lipid-lowering therapies.30,31 As a consequence, European Society of Cardiology guidelines warrant caution when interpreting CAC score values among patients on lipid-lowering therapy.11 Our finding is reassuring, by showing that the prognostic implications of CAC score remain valid among patients already on lipid-lowering treatment. This observation is consistent with a recent analysis from the CAC consortium, which showed that CAC retains robust risk prediction in statin users, though with a slightly weaker power compared with a statin nonuser, likely explained by the changing relationship of CAC density among statin users.32

The findings of this meta-analysis should be interpreted in the context of some limitations. First, this is a study-level meta-analysis, and the findings provide mean study-level effects. Second, the rate of crossovers and the variable exposures to lipid-lowering drugs in the studies included in the analysis may complicate the interpretation of the results. Third, we report both unadjusted and adjusted pooled effect estimates, as the latter were available for only 2 studies (and 1 study for the CAC 0 group). Moreover, despite adjustment, this analysis cannot account for unmeasured covariates and do not completely eliminate confounding bias. However, the consistency between the results of the analyses support the validity of our observations. Fourth, ASCVD definition varied among studies. Even though the effect size of benefit of lipid-lowering therapy was similar among individual cardiovascular outcomes (with slight attenuation for stroke and cardiovascular death, compared with myocardial infarction and coronary revascularization)1, the reported relative effect estimates should be interpreted in this context.