Author + information
- Received December 6, 2016
- Revision received January 18, 2017
- Accepted January 25, 2017
- Published online February 5, 2018.
- Martin Bødtker Mortensen, MD, PhDa,∗ (, )
- Erling Falk, MD, DMSca,
- Dong Li, MDb,
- Khurram Nasir, MD, MPHc,d,e,
- Michael J. Blaha, MD, MPHe,
- Veit Sandfort, MDf,
- Carlos Jose Rodriguez, MD, MPHg,
- Pamela Ouyang, MD, MBBSh and
- Matthew Budoff, MDb
- aDepartment of Cardiology, Aarhus University Hospital, Aarhus, Denmark
- bDivision of Cardiology, Los Angeles Biomedical Research Institute at Harbor-UCLA, University of California Los Angeles, Torrance, California
- cCenter for Healthcare Advancement and Outcomes, Baptist Health South Florida, Miami, Florida
- dDepartment of Epidemiology, Robert Stempel College of Public Health and Department of Medicine, Herbert Wertheim College of Medicine, Florida International University, Florida
- eThe Johns Hopkins Ciccarone Center for the Prevention of Heart Disease, Baltimore, Maryland
- fRadiology and Imaging Sciences, National Institutes of Health, Bethesda, Maryland
- gDivision of Public Health Sciences, Department of Epidemiology and Prevention, Wake Forest School of Medicine, Winston-Salem, North Carolina
- hThe John Hopkins University School of Medicine, Baltimore, Maryland
- ↵∗Address for correspondence:
Dr. Martin Bødtker Mortensen, Department of Cardiology, Aarhus University Hospital, Palle Juul Jensens Boulevard 99, Aarhus N, Jylland 8200, Denmark.
Objectives This study sought to determine whether coronary artery calcium (CAC) could be used to optimize statin allocation among individuals for whom trial-based evidence supports efficacy of statin therapy.
Background Recently, allocation of statins was proposed for primary prevention of atherosclerotic cardiovascular disease (ASCVD) based on proven efficacy from randomized controlled trials (RCTs) of statin therapy, a so-called trial-based approach.
Methods The study used data from MESA (Multi-Ethnic Study of Atherosclerosis) with 5,600 men and women, 45 to 84 years of age, and free of clinical ASCVD, lipid-lowering therapy, or missing information for risk factors at baseline examination.
Results During 10 years’ follow-up, 354 ASCVD and 219 hard coronary heart disease (CHD) events occurred. Based on enrollment criteria for 7 RCTs of statin therapy in primary prevention, 73% of MESA participants (91% of those >55 years of age) were eligible for statin therapy according to a trial-based approach. Among those individuals, CAC = 0 was common (44%) and was associated with low rates of ASCVD and CHD (3.9 and 1.7, respectively, per 1,000 person-years). There was a graded increase in event rates with increasing CAC score, and in individuals with CAC >100 (27% of participants) the rates of ASCVD and CHD were 18.9 and 12.7, respectively. Consequently, the estimated number needed to treat (NNT) in 10 years to prevent 1 event varied greatly according to CAC score. For ASCVD events, the NNT was 87 for CAC = 0 and 19 for CAC >100. For CHD events, the NNT was 197 for CAC = 0 and 28 for CAC >100.
Conclusions Most MESA participants qualified for trial-based primary prevention with statins. Among the individuals for whom trial-based evidence supports efficacy of statin therapy, CAC = 0 and CAC >100 were common and associated with low and high cardiovascular risks, respectively. This information may guide shared decision making aimed at targeting evidence-based statins to those who are likely to benefit the most.
Low-density lipoprotein cholesterol lowering by using HMG-CoA (3-hydroxy-3-methylglutaryl-coenzyme A) reductase inhibitors, also known as statins, constitutes the cornerstone of pharmacological prevention of atherosclerotic cardiovascular disease (ASCVD). Although it is widely accepted that statins should be offered to patients with clinical ASCVD (secondary prevention), controversies exist about whom to treat for primary prevention. Leading international guidelines for the use of ASCVD prevention agree on the principle of allocating statin therapy based on absolute 10-year risk estimates of future ASCVD (1–3). This long-held principle, however, was recently questioned by leading cardiovascular investigators who proposed a paradigm shift in ASCVD prevention in which statin eligibility is based on randomized controlled trials (RCT) of statin therapy (4–7). In this alternative proposal, allocation of statins is based strictly on proven trial evidence (“trial-based approach”), that is, on the principle of “what works” and “in whom.” The rationale behind such a trial-based approach is clear: no RCTs of statin therapy have ever enrolled participants based on 10-year ASCVD risk assessment (the approach recommended for statin allocation by current guidelines), and abundant data from large-scale RCTs have now proven the efficacy and safety of statin therapy in a wide range of different patient populations. Unfortunately, as recently highlighted (8), most individuals eligible for statin therapy using a trial-based approach are at low absolute risk of ASCVD, in whom the net benefit of treatment may be questioned.
Nevertheless, accepting the rationale behind a trial-based approach to statin therapy, we sought to determine whether assessment of subclinical atherosclerosis, the root cause of ASCVD, could be used to improve trial-based statin allocation. Specifically, we hypothesized that assessment of coronary artery calcium (CAC) among individuals for whom trial-based evidence supported efficacy of statin therapy could be used to identify subgroups with high and low ASCVD event rates and, thereby, those individuals who could be expected to benefit the most, and least, from trial-based evidence supporting primary prevention with statin therapy.
MESA (Multi-Ethnic Study of Atherosclerosis) is a National Institutes of Health/National Heart, Lung, and Blood Institute-funded study of the characteristics of subclinical atherosclerosis and is designed to identify risk factors involved in progression of atherosclerosis to clinical ASCVD. A total of 6,814 men and women, 45 to 84 years of age and free of clinical ASCVD at baseline examination, were recruited between July 2000 and September 2002. Subjects were enrolled at 6 sites in the United States (Baltimore, Maryland; Chicago, Illinois; Forsyth County, North Carolina; Los Angeles, California; New York City, New York; and St. Paul, Minnesota). Details on the design and organization were published previously (9).
Risk factor determination and assessment
The baseline examination in MESA included completion of an interview/questionnaire, a physical examination, and blood sampling for biochemical measurements. In the interview, MESA staff collected information on conventional and nonconventional risk factors. Systolic and diastolic blood pressures were measured at rest by using an automated oscillometric sphygmomanometer (Dinamap Pro 1000, Critikon, Tampa, Florida), using the mean of the previous 2 measurements for analysis. Blood samples were drawn after a 12-h fast and used for the measurement of total cholesterol, LDL-C, and triglycerides at the collaborative Studies Clinical Laboratory (Fairview University Medical Center, Minneapolis, Minnesota). Smoking was defined as current smoking by self-report. Diabetes was defined as self-reported diabetes, a fasting glucose concentration of ≥7.0 mmol/l or use of antidiabetic drugs.
CAC score measurements
All MESA participants underwent noncontrast cardiac-gated computed tomography at baseline examination to determine the Agatston CAC score. Participants were scanned twice, using mean CAC score for analysis. The estimated average radiation dose was 0.89 mSv.
Trial-based recommendations for statin therapy
A trial-based approach to statin therapy for primary prevention based on currently available evidence is guided by enrollment criteria in the following 7 large RCTs (named in chronological order by publication year): WOSCOPS (West of Scotland Coronary Prevention Study) (10), AFCAPS/TexCAPS (Air Force/Texas Coronary Atherosclerosis Prevention Study) (11), ASCOT-LLA (Anglo-Scandinavian Cardiac Outcomes Trial-Lipid Lowering Arm) (12), CARDS (Collaborative Atorvastatin Diabetes Study) (13), MEGA (Management of Elevated Cholesterol in the Primary Prevention Group of Adult Japanese) (14), JUPITER (Justification for the Use of Statins in prevention: An Intervention Trial Evaluating Rosuvastatin) (15), and HOPE-3 (Heart Outcomes Prevention Evaluation-3) (16). Characteristics of these 7 RCTs to guide trial-based allocation of statins in primary prevention of ASCVD are shown in Figure 1.
Cardiovascular disease endpoints
Ascertainment of events has been described previously and is available at the MESA Website (17). Briefly, at intervals of 9 to 12 months, trained MESA personnel contacted participants or family members to inquire about ASCVD diagnosis, including hospital admissions, outpatient diagnoses, and deaths. Follow-up was completed in 92% of living participants. Medical records were obtained for approximately 98% of hospital admissions and 95% of outpatient diagnoses. A MESA study committee, including cardiologists, neurologists, and epidemiologists, adjudicated every event.
For this study, we defined coronary heart disease (CHD) events as myocardial infarction, resuscitated cardiac arrest, and CHD death. ASCVD was defined as CHD plus fatal and nonfatal strokes. Myocardial infarction was diagnosed based on the combination of symptoms, electrocardiographic findings, and levels of cardiac biomarkers. Hospital records and family interviews were used to determine whether a death was related to CHD. Stroke was diagnosed based on a documented focal neurological deficit lasting 24 h or until death or, if <24 h, with imaging evidence of relevant brain lesions. In this study, participants were followed for 10 years (i.e., data were ended at 10 years).
Baseline characteristics are presented as proportions for categorical variables and as median (interquartile range) for continuous variables.
We calculated the number and percentage of participants eligible for statin therapy under the above-described trial-based approach. Among these trial-based eligible individuals, for whom RCT evidence supports efficacy of statin therapy, we assessed the distribution of CAC using 3 well-defined CAC groups: 0, 1 to 100, and >100 (18–21).
To determine whether CAC could be used to risk stratify trial-based eligible individuals, we calculated the 10-year ASCVD and CHD event rates across the 3 CAC groups and used Cox regression modeling (analyzing time to event) to obtain multivariate-adjusted hazard ratios. Analyses were adjusted for race and MESA site. Furthermore, we used Kaplan-Meier estimates to describe the occurrences of ASCVD and CHD events over time, stratified by the CAC groups.
Finally, we calculated a 10-year number needed to treat (NNT10) to prevent 1 ASCVD or CHD event by assuming a 30% relative risk reduction with statin therapy in primary prevention (22,23). The NNT10 was calculated for each CAC group as the reciprocal of the absolute risk difference in 10-year event rates. In a sensitivity analysis, we also estimated 5-year NNT (NNT5), using 5-year Kaplan-Meier estimates (24) to better comply with follow-up length in the RCTs. Furthermore, in a second sensitivity analysis, we recalculated NNT10 after assuming a more optimistic benefit of long-term statin therapy for primary prevention (40% relative risk reduction). Analyses were performed using Stata version 13.1 SE (Stata Corp., College Station, Texas).
A total of 6,814 men and women were included in MESA. After exclusion of individuals taking lipid-lowering medication (n = 1,100) or with missing information (n = 114), 5,600 individuals were available for this study. Baseline characteristic of the study population are shown in Table 1. Median age was 61 years, and 53% were women.
Statin eligibility based on randomized statin trials
Based on enrollment criteria used in 7 high-quality RCTs of statin therapy (trial-based approach), 4,085 individuals (73%) were eligible for primary prevention using statins (Figure 2). More men than women met enrollment criteria (82% vs. 65%, respectively). Notably, among those >55 years of age, 91% qualified for trial-based statin therapy (Online Figure 1). Baseline characteristics of participants stratified by statin eligibility are presented in Table 2. Individuals eligible for statins were older and had a higher burden of cardiovascular risk factors, including higher blood pressure and more atherogenic lipid profile than individuals who did not fulfill enrollment criteria in any RCT of statin therapy. When we assessed statin eligibility by each trial individually, 55% of MESA participants qualified for statin therapy based on HOPE-3 trial alone. For comparison, statin eligibility varied from 1% (WOSCOPS) to 18% (JUPITER) in the other 6 trials (Online Figure 2, Online Table 1). More than one-half of statin-eligible individuals met enrollment criteria in 2 or more of the 7 RCTs.
CAC distribution among statin eligible individuals
Among the 4,085 individuals meeting enrollment criteria in RCTs of statin therapy, nearly one-half had no detectable CAC (CAC = 0), and more than one-fourth had CAC >100 (Figure 3). The burden of CAC differed in a sex-specific manner. The CAC = 0 status was more common among women than men; the opposite was the case for CAC >100. Overall, the number needed to screen (NNS) to identify 1 person with CAC = 0 was 2.3, and the NNS to find 1 with CAC >100 was 3.7 (Table 3). The NNS to identify 1 person with either CAC = 0 or CAC >100 was <2. The distribution of CAC among individuals meeting enrollment criteria in each of the 7 statin trials varied considerably. Using MEGA criteria would have included the most individuals (59%) with no CAC, whereas using WOSCOPS criteria would have included the fewest (38%) with no CAC (Online Figure 3).
Clinical events in statin-eligible individuals
Among those who qualified for trial-based statin therapy, we observed 332 ASCVD events during 10-year follow-up, of which 208 were CHD events (Table 4). Kaplan-Meier cumulative-event curves for ASCVD and CHD by CAC score (0, 1 to 100, and >100) are shown in Figure 4. There was a strong relationship between the burden of CAC and both ASCVD and CHD events (Table 4, Online Table 2). Among individuals with no CAC, the event rates were low (3.9 for ASCVD and 1.7 for CHD per 1,000 person-years), whereas the event rates were considerably higher in individuals with CAC >100 (18.9 for ASCVD and 12.7 for CHD per 1,000 person-years). Compared with a CAC score of 0, CAC scores above 100 were associated with an adjusted hazard ratio of 5.2 (95% confidence interval [CI]: 3.9 to 6.4) for ASCVD and 8.0 (95% CI: 5.3 to 12.1) for CHD. Interestingly, event rates were similar in men and women separately when stratified by CAC score (Table 4).
NNT stratified by CAC group
Assuming a 30% relative risk reduction with statin therapy, the NNT10 to prevent 1 ASCVD event among individuals meeting enrollment criteria in RCTs of statin therapy was 87 for those with CAC = 0 compared with 19 for those with CAC >100 (Figure 5). For CHD events, the NNT10 to prevent 1 event was 197 for those without CAC compared to 28 for those with CAC >100. For ASCVD events there was no sex difference, but women with CAC = 0 had a higher NNT to prevent 1 CHD event than men. As most statin trials have mean follow-up time of ≤5 years, we also calculated NNT for 5 years (NNT5) (Online Figure 4). The NNT5 ranged from 194 (CAC = 0) to 40 (CAC >100) for ASCVD and from 725 (CAC = 0) to 54 (CAC >100) for CHD events. Notably, among women with CAC = 0 (54% of all women), the estimated NNT5 to prevent 1 CHD event was >1,000. In a secondary sensitivity analysis, we recalculated the NNT10 assuming 40% relative risk reduction by long-term statin therapy. In this analysis, the NNT10 was as low as 14 for ASCVD events and 21 for CHD events in those with CAC >100 (Online Figure 5).
Among MESA participants, as many as 73% met enrollment criteria in 1 or more of 7 high-quality RCTs of statin therapy for primary prevention and would therefore be eligible for trial-based statin therapy. In these trial-based eligible individuals, the burden of CAC differed substantially with nearly one-half having CAC = 0 and, thus, very low event rates. In those with CAC = 0, the NNT to prevent 1 event (ASCVD or CHD) was unfavorably high, with an NNT10 of 87 for ASCVD and 197 for CHD. In contrast, in the large subpopulation with CAC score >100, event rates were much higher and associated with considerably more favorable NNT10 (19 for ASCVD and 28 for CHD). Hence, for health care providers who prefer a trial-based approach to primary prevention with statins, knowing the CAC score may help targeting of treatment to those at highest risk for ASCVD and, thus, to those who are likely to benefit the most from statin therapy.
Trial-based statin therapy for primary prevention: is it reasonable to treat all?
In 1995, the first larger RCT of statin use in primary prevention (WOSCOPS) (10) was published in which the efficacy of statin therapy was documented in a selected subgroup of high-risk men with hypercholesterolemia. Only 1% of MESA participants met enrollment criteria for WOSCOPS. Since then, primary prevention with statins has proven effective in other carefully selected individuals with specific risk-factor profiles, progressively expanding the indication for primary prevention with statins. Based on enrollment criteria used in the first 6 of the 7 RCTs (10–15), 52% of MESA participants were eligible for primary prevention with statin therapy. Similar results were recently reported from 2 European population-based cohort studies, the Copenhagen General Population Study (56%) (8) and the Rotterdam Study (53%) (25).
In 2016, RCT evidence for primary prevention with statins became stronger and more inclusive, with publication of the HOPE-3 trial (16). This pragmatic trial had enrolled “intermediate-risk” persons without known ASCVD in whom clear trial-based evidence for efficacy of statins was still lacking. The HOPE-3 trial provided important missing evidence, and 55% of MESA participants were statin-eligible based solely on enrollment criteria for this trial. Considering not only HOPE-3 but the totality of evidence from all 7 randomized statin trials, 73% of MESA participants 40 to 84 years of age qualified for trial-based primary prevention with statins, increasing to 91% when considering only those >55 years of age. This proportion would most likely be even higher in a real-world population (26).
Hence, after publication of the statin arm of HOPE-3, near-complete RCT evidence has now been provided for a universal, pragmatic approach to primary prevention of ASCVD based on a fixed low-to-moderate statin dose from 55 years of age, a provocative concept introduced by Wald and Law (27) in 2003. However, a critical question is whether such a pragmatic approach is reasonable in countries where a simple test is available that could distinguish between those who need and those who do not need to take a statin pill every day for the rest of their lives.
Precision medicine: CAC to guide trial-based statin allocation
Over the short term, people without atherosclerosis are at low risk for ASCVD and the higher the burden of atherosclerosis, the higher the risk for ASCVD. Although CAC is not a marker of the earliest coronary atherosclerotic lesions, those with CAC = 0 are at very low risk for ASCVD and mortality for up to 15 years (18,21,28–33). At the other end of the risk spectrum, those with CAC >100 have a risk for a first ASCVD event that approaches that seen for a recurrent event in patients with established ASCVD (secondary prevention) (19). Thus, in primary prevention, it makes sense to identify those with CAC = 0 to avoid overtreatment and those with CAC >100 to avoid undertreatment and ensure long-term adherence to a cost-effective treatment (34).
In the present study, we confirmed that CAC = 0 at baseline examination was associated with very low CHD and ASCVD event rates for at least 10 years, known as the power of zero (35). Of course, to prevent events, there need to be events to prevent (19). Nearly one-half of MESA participants who were eligible for trial-based statin therapy had CAC = 0. In that low-risk population (CAC = 0), the NNT5 to prevent 1 CHD event was high (>500 in men and >1,000 in women, assuming 30% event reduction with statin therapy). Screening just 3 persons who were eligible for trial-based statin therapy would identify 1 who did not need this treatment. At the other end of the risk spectrum, more than one-fourth of MESA participants had CAC >100 and a high 10-year event rate. Most events occurred in this high-risk subpopulation. Overall, just 2 persons need to be screened to find just 1 with either CAC = 0 (do not treat) or CAC >100 (treat and ensure long-term adherence).
The size of the study population and length of follow-up allowed us to assess the benefit of CAC assessment in men and women separately. The NNS to find 1 person with CAC = 0 or CAC >100 in men was similar to that in women (<2), but more women than men had CAC = 0, whereas the opposite was the case for CAC >100. Notably, the 10-year cumulative ASCVD risk was >15% in both men and women with CAC >100, that is, far above the current 7.5% 10-year ASCVD risk threshold for statin therapy identified by the American College of Cardiology/American Heart Association (ACC/AHA) guidelines. Thus, as the ACC/AHA guidelines currently recommend that statin therapy might be considered in selected individuals with CAC ≥300 (Class IIb recommendation), our results suggest that this cut point may reasonably be lowered to 100 in future guidelines. Given that the price for CAC testing is now low (U.S. ≈$100), the additional information on ASCVD risk that CAC provides (especially when treatment decision is uncertain) may be worth the expense and, under some circumstances, even be cost-effective (36). In patients with detectable CAC, knowing this may increase adherence to preventive medication (37), which may further improve ASCVD outcome (38). These considerations are obvious topics for evidence-based and meaningful patient-physician discussions on initiation of statin therapy for primary prevention (39).
First, our ability to consider all exclusion criteria used in the randomized statin trials was limited. However, potential exclusion criteria were not mentioned in the trial-based proposal and are often ignored in routine clinical practices (40). Second, although we excluded individuals taking lipid-lowering medication at baseline examination, MESA participants were informed about their CAC scores, which may have led to selective uptake of preventive measures (including statin) among individuals with high CAC scores that could have influenced event rates. However, this would be expected to weaken the association among CAC and ASCVD and CHD events and, thus, cannot explain our results. Third, we assumed a 30% relative risk reduction with statin therapy based on Cochrane analyses (22), although that may have varied according to both treatment time and dose/type of statin. Fourth, most statin trials have a follow-up of ≤5 years. However, using 5 years of follow-up instead of 10 years did not affect the main results or conclusion.
Strengths of our study include the high-quality assessment of risk factors at baseline (enabling assessment of enrollment criteria in the 7 RCTs), adjudicated events over 10 years of follow-up and the size of the study population that allowed sex-specific assessment of the trial-based approach to statin therapy in a modern, multiethnic population.
After the HOPE-3 trial, most of the middle-aged and elderly MESA participants free of ASCVD would meet enrollment criteria used in at least 1 randomized statin trial. Evidence from RTCs now supports primary prevention with statins in nearly all men and women >55 years of age. However, nearly one-half of those considered statin-eligible based on RCTs had CAC = 0 and a very low event rate, and one-fourth had CAC >100 and a high event rate. For health care providers and patients who are reluctant to enter into treatment of all with statins from age 55 years, this information may help in shared decision making aimed at targeting prevention to those at highest risk.
COMPETENCY IN MEDICAL KNOWLEDGE: Among individuals with trial evidence supporting statin efficacy, quantification of subclinical atherosclerosis using the Agatston CAC score can be used to identify individuals with questionable (if CAC = 0) and substantial (if CAC >100) benefit of statin therapy.
TRANSLATIONAL OUTLOOK: Future research is needed to evaluate how best to incorporate CAC-guided allocation of statin therapy in routine clinical practice.
The authors thank the investigators, the staff, and the participants of the MESA study for their valuable contributions. A complete list of participating MESA investigators and institutions can be found at the MESA Website.
This research was supported by National Heart, Lung, and Blood Institute contracts HHSN268201500003I, N01-HC-95159, N01-HC-95160, N01-HC-95161, N01-HC-95162, N01-HC-95163, N01-HC-95164, N01-HC-95165, N01-HC-95166, N01-HC-95167, N01-HC-95168, and N01-HC-95169; and The National Center for Research Resources (NCRR) grants UL1-TR-000040 and UL1-TR-001079. Dr. Nasir is a consultant for Regeneron; a member of the Quest Diagnostic advisory board; and has received honoraria from Regeneron and Quest Diagnostic. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- atherosclerotic cardiovascular disease
- coronary artery calcium
- pooled cohort equation
- randomized controlled trial
- Received December 6, 2016.
- Revision received January 18, 2017.
- Accepted January 25, 2017.
- 2018 American College of Cardiology Foundation
- Stone N.J.,
- Robinson J.G.,
- Lichtenstein A.H.,
- et al.
- JBS3 Board
- Ridker P.M.,
- Rose L.,
- Cook N.R.
- Ridker P.M.
- Mortensen M.B.,
- Afzal S.,
- Nordestgaard B.G.,
- Falk E.
- Sever P.S.,
- Dahlöf B.,
- Poulter N.R.,
- et al.
- ↵MESA Coordinating Center. MESA website. 2015. Available at: http://www.mesa-nhlbi.org. Accessed May 24, 2017.
- Martin S.S.,
- Blaha M.J.,
- Blankstein R.,
- et al.
- Bittencourt M.S.,
- Blaha M.J.,
- Blankstein R.,
- et al.
- Nasir K.,
- Bittencourt M.S.,
- Blaha M.J.,
- et al.
- Ebrahim S.,
- Taylor F.C.,
- Brindle P.
- Ridker P.M.,
- MacFadyen J.G.,
- Fonseca F.A.H.,
- et al.
- Pavlovic J.,
- Greenland P.,
- Deckers J.W.,
- et al.
- Wald N.J.,
- Law M.R.
- Hecht H.S.
- Greenland P.
- Valenti V.,
- Ó Hartaigh B.,
- Heo R.,
- et al.
- Mortensen M.B.,
- Fuster V.,
- Muntendam P.,
- et al.
- Blaha M.,
- Budoff M.J.,
- Shaw L.J.,
- et al.
- Rossello X.,
- Pocock S.J.,
- Julian D.G.
- Blaha M.J.,
- Silverman M.G.,
- Budoff M.J.
- Pletcher M.J.,
- Pignone M.,
- Earnshaw S.,
- et al.
- Bansilal S.,
- Castellano J.M.,
- Garrido E.,
- et al.
- Martin S.S.,
- Sperling L.S.,
- Blaha M.J.,
- et al.
- Pocock S.J.,
- Gersh B.J.