Anda di halaman 1dari 11

JOURNAL OF THE AMERICAN COLLEGE OF CARDIOLOGY VOL. 68, NO.

9, 2016

2016 BY THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION. PUBLISHED BY ISSN 0735-1097

ELSEVIER. THIS IS AN OPEN ACCESS ARTICLE UNDER THE CC BY LICENSE http://dx.doi.org/10.1016/j.jacc.2016.05.084

(http://creativecommons.org/licenses/by/4.0/).

ORIGINAL INVESTIGATIONS

A Simple Disease-Guided Approach to


Personalize ACC/AHA-Recommended
Statin Allocation in Elderly People
The BioImage Study

Martin Bdtker Mortensen, MD, PHD,a Valentin Fuster, MD, PHD,b Pieter Muntendam, MD,c Roxana Mehran, MD,b
Usman Baber, MD, MS,b Samantha Sartori, PHD,b Erling Falk, MD, DMSCa

ABSTRACT

BACKGROUND The 2013 American College of Cardiology (ACC)/American Heart Association (AHA) guidelines recom-
mend primary prevention with statins for individuals with $7.5% 10-year risk for atherosclerotic cardiovascular disease
(ASCVD). Everyone living long enough will become eligible for risk-based statin therapy due to age alone.

OBJECTIVES This study sought to personalize ACC/AHA risk-based statin eligibility using noninvasive assessment of
subclinical atherosclerosis.

METHODS In 5,805 BioImage participants without known ASCVD at baseline, those with $7.5% 10-year ASCVD risk
were down-classied from statin eligible to ineligible if imaging revealed no coronary artery calcium (CAC) or carotid
plaque burden (cPB). Intermediate-risk individuals were up-classied from optional to clear statin eligibility if CAC
was $100 (or equivalent cPB).

RESULTS At a median follow-up of 2.7 years, 91 patients had coronary heart disease and 138 had experienced a cardio-
vascular disease event. Mean age of the participants was 69 years, and 86% qualied for ACC/AHA risk-based statin therapy,
with high sensitivity (96%) but low specicity (15%). CAC or cPB scores of 0 were common (32% and 23%, respectively) and
were associated with low event rates. With CAC-guided reclassication, specicity for coronary heart disease events improved
22% (p < 0.0001) without any signicant loss in sensitivity, yielding a binary net reclassication index (NRI) of 0.20
(p < 0.0001). With cPB-guided reclassication, specicity improved 16% (p < 0.0001) with a minor loss in sensitivity (7%),
yielding an NRI of 0.09 (p 0.001). For cardiovascular disease events, the NRI was 0.14 (CAC-guided) and 0.06
(cPB-guided). The positive NRIs were driven primarily by down-classifying the large subpopulation with CAC 0 or cPB 0.

CONCLUSIONS Withholding statins in individuals without CAC or carotid plaque could spare a signicant proportion
of elderly people from taking a pill that would benet only a few. This individualized disease-guided approach is
simple and easy to implement in routine clinical practice. (J Am Coll Cardiol 2016;68:88191) 2016 by the American
College of Cardiology Foundation. Published by Elsevier. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).

Listen to this manuscripts From the aDepartment of Cardiology, Aarhus University Hospital, Aarhus, Denmark; bCardiovascular Institute, Mount Sinai
audio summary by Medical Center, Icahn School of Medicine at Mount Sinai, New York, New York; and cscPharmaceuticals, Lexington, Massachu-
JACC Editor-in-Chief setts. The BioImage Study was designed by the High-Risk Plaque Initiative, a pre-competitive industry collaboration funded by BG
Dr. Valentin Fuster. Medicine, Abbott Vascular, AstraZeneca, Merck & Co., Philips, and Takeda. Dr. Mehran has received research grant support from
Eli Lilly/Daiichi-Sankyo, Bristol-Myers Squibb, AstraZeneca, The Medicines Company, and OrbusNeich; has served as a consultant
for Janssen Pharmaceuticals, Osprey Medical, Watermark Research Partners, and Medscape; and has served on the scientic
advisory board of Abbott Laboratories. All other authors have reported that they have no relationships relevant to the contents of
this paper to disclose. Tasneem Naqvi, MD, served as Guest Editor for this paper.

Manuscript received March 15, 2016; revised manuscript received May 17, 2016, accepted May 26, 2016.
882 Mortensen et al. JACC VOL. 68, NO. 9, 2016

Disease-Guided Initiation of Statin Therapy AUGUST 30, 2016:88191

I
ABBREVIATIONS n 2013, the American College of Cardiol- of age without known ASCVD at baseline examination
AND ACRONYMS ogy (ACC) and the American Heart Asso- between January 2008 and June 2009. The BioImage
ciation (AHA) released new guidelines cohort is sex-balanced and included racial/ethnic
ACC = American College of
Cardiology
on risk assessment (1) and cholesterol treat- minorities corresponding to the overall U.S. popula-
ment (2). For use in primary prevention of tion. The primary objective was to identify predictive
AHA = American Heart
Association atherosclerotic cardiovascular disease biomarkers for near-term ASCVD events in the
ASCVD = atherosclerotic (ASCVD), a new risk model on the basis of noninvasive imaging group (n 6,102) (12). The study
cardiovascular disease the Pooled Cohort Equations (PCE) was intro- was approved by the Western Institutional Review
CAC = coronary artery calcium duced (1), together with an online ASCVD risk Board, Olympia, Washington. Informed consent and
CHD = coronary heart disease calculator (3). The indication for statin ther- Health Insurance Portability and Accountability
cPB = carotid plaque burden apy in individuals without ASCVD or diabetes Act authorization were obtained from all study
CVD = cardiovascular disease
was expanded by lowering the threshold for participants.
treatment to $7.5% 10-year ASCVD risk esti- Baseline examination included assessment of car-
MI = myocardial infarction
mated by the PCE-based risk calculator (Class diovascular risk factors and screening for subclinical
PCE = Pooled Cohort Equations
I recommendation) (2). That threshold to (asymptomatic) atherosclerosis as previously
initiate statin therapy was supported by risk-benet described (12) and specied in the Online Appendix.
and cost-effectiveness analyses (1,2,4). All study participants underwent noncontrast
computed tomography scanning to determine the
SEE PAGE 892
Agatston coronary artery calcium (CAC) score and
carotid ultrasound imaging to detect and quantify
With the growing elderly population, the dominant
carotid plaque. By using a novel sweep method,
effect of age on eligibility for statin therapy might need
described previously (13) and in the Online Appendix,
reconsideration (57). All healthy people with optimal
all cross-sectional plaque areas were summed as ca-
cardiovascular risk factors will automatically, due to
rotid plaque burden (cPB).
age-related risk alone, pass the 7.5% 10-year ASCVD
risk threshold and qualify for ACC/AHA-recommended DISEASE-GUIDED RECLASSIFICATION. The PCEs are

statin therapy between age 63 and 71 years (depending applicable to ASCVD-free individuals 40 to 79 years of
on sex and ethnicity). This universal eligibility for age (1). The ACC/AHA guidelines (2) dene 4 statin-
statin therapy with aging will inevitably lead to over- benet groups; the following 3 are relevant for pri-
treatment of many older individuals who do not have mary prevention of ASCVD (Class 1 recommendation):
the disease (atherosclerosis) that the treatment is 1. Individuals without ASCVD or diabetes, 40 to 75
intended to prevent or stabilize. Another related years of age, with a low-density lipoprotein
concern: PCEs tend to overestimate 10-year ASCVD cholesterol (LDL-C) of 70 to 189 mg/dl, and with an
risk in contemporary lower-risk populations, which estimated 10-year ASCVD risk of $7.5%.
also may lead to overtreatment (2,810). 2. Individuals with diabetes, 40 to 75 years of age,
The purpose of the present study was to evaluate a and LDL-C 70 to 189 mg/dl.
more personalized approach to primary prevention 3. Individuals with primary elevations of LDL-C
with statins by adding a simple disease-guided $190 mg/dl
reclassication step after formal ACC/AHA-
recommended risk assessment. Thus, the decision to Additionally, statin therapy is reasonable when the
treat with statins is simply guided by the absence (do 10-year ASCVD risk is 5% to <7.5% (Class IIa recom-
not treat) or presence (treat) of subclinical athero- mendation) and might be considered at even lower
sclerosis detected by noninvasive imaging of the cor- risk and beyond the age range of 40 to 75 years (Class
onary and carotid arteries (Central Illustration). We IIb recommendation) (2).
tested this practical, disease-guided reclassication Accepting these recommended cut-points for pri-
approach to statin allocation for primary prevention of mary prevention with statins as a useful starting
ASCVD in the contemporary, multiethnic BioImage point, we tested the following disease-guided
cohort. reclassication approach:

METHODS  To improve specicity (less overtreatment), in-


dividuals with $7.5% 10-year ASCVD risk estimated
The design and objectives of the BioImage Study by PCE are down-classied from statin eligible to
(NCT00738725) have been published previously (11). ineligible if CAC or cPB is absent.
BioImage was a prospective observational cohort of  To improve sensitivity (less undertreatment), in-
men 55 to 80 years of age and women 60 to 80 years dividuals with intermediate ASCVD risk (5%
JACC VOL. 68, NO. 9, 2016 Mortensen et al. 883
AUGUST 30, 2016:88191 Disease-Guided Initiation of Statin Therapy

C ENTR AL I LL U STRA T I O N Disease-Guided Primary Prevention With Statins

Mortensen, M.B. et al. J Am Coll Cardiol. 2016;68(9):88191.

(A) Following guideline-recommended formal risk assessment and atherosclerosis imaging, statin-eligible individuals are down-classied to ineligible in the absence of
atherosclerosis, and statin-ineligible individuals are up-classied to eligible if signicant atherosclerosis is present. This principle facilitates an informed clinician-patient
discussion, leading to an individualized treatment decision. (B) In the BioImage cohort, benecial reclassication was achieved using coronary artery calcium (CAC) and
carotid plaque burden (cPB). ACC/AHA American College of Cardiology/American Heart Association; ASCVD atherosclerotic cardiovascular disease; CHD coronary
heart disease; CVD cardiovascular disease; NRI net reclassication index.

to <7.5% estimated by PCE) are up-classied from disease (CHD) to ASCVD (CHD stroke) using the
optional to statin eligible if the CAC is $100 or cPB PCE-based risk calculator, and the threshold for statin
is $300 mm 2 (equivalent to CAC $100). therapy was lowered (2). Consequently, the evidence
base for the clinical utility of atherosclerosis imaging
CLINICAL OUTCOMES. The 2013 ACC/AHA guidelines in risk assessment under the previous ACC/AHA
changed the predicted outcome from coronary heart guidelines (14) was no longer applicable and needed
884 Mortensen et al. JACC VOL. 68, NO. 9, 2016

Disease-Guided Initiation of Statin Therapy AUGUST 30, 2016:88191

to be updated (1,2). To inform this revision, we pre- time according to zero, middle, and top groups of CAC
sent results separately for the following outcomes: and cPB.
1) CHD (spontaneous myocardial infarction [MI], un- The clinical usefulness of assessing subclinical
stable angina [UA], and coronary revascularization); atherosclerosis in addition to the ACC/AHA guidelines
and 2) cardiovascular disease (CVD) (CHD ischemic depends on the ability to correctly reclassify in-
stroke, and cardiovascular death). An independent dividuals across decision thresholds. We therefore
clinical events committee used source medical re- calculated the sensitivity, specicity, and binary net
cords to adjudicate CHD and CVD events as described reclassication index (NRI) after applying the
in the Online Appendix. disease-guided reclassication described in the pre-
STATISTICAL APPROACH. Baseline characteristics of vious text. In the rst analysis, only participants with
BioImage participants were analyzed according to 10- a 10-year ASCVD risk of $7.5% to 15% could be down-
year ASCVD risk estimated using the PCEs underlying classied from statin eligible to ineligible by absence
the ACC/AHA online risk calculator (13). To compare of CAC or cPB. In the second analysis, all participants
the clinical performance of CAC and cPB, we created 3 with a 10-year ASCVD risk of $7.5% could be down-
comparable groups (zero, middle, and top) for each classied to statin ineligible. The binary NRI (to
modality, with a similar proportion of BioImage par- treat or not to treat) is the sum of D sensitivity and
ticipants in the 2 top groups. The top CAC group Dspecicity and can be in the range of 2 to 2.
comprised those with CAC $100, and the top cPB
group comprised those with cPB $300 mm 2 (chosen RESULTS
to match the percentile for CAC 100, giving 2 top
groups of similar size but selected differently). The In the BioImage study, 6,102 participants underwent
zero groups were absence of CAC (CAC 0) and cPB noninvasive imaging to assess subclinical athero-
(cPB 0), respectively. We then calculated the sclerosis. Due to missing data, 297 participants were
number needed to screen (NNS) to identify 1 person excluded, yielding a nal study population of 5,805
belonging to the zero or top group of subclinical adults.
atherosclerosis by dividing 100 by the percentage of Per baseline characteristics (Table 1), the study
people with the given amount of disease. populations mean age was 69 years, and 56% of
The association (hazard ratio [HR]) of CAC and cPB participants were women. The great majority (86%) of
groups with development of CHD and CVD was the BioImage cohort was statin eligible because of an
assessed using Cox regression models analyzing time estimated 10-year ASCVD risk $7.5%; risk was $15%
to event. Analyses were multivariable and adjusted in 55% of participants (Central Illustration).
for baseline characteristics. We used Kaplan-Meier According to the relationship between cardiovas-
estimates of cumulative incidence to compare (log- cular risk factors and subclinical atherosclerosis at
rank test) the occurrence of CHD and CVD events over baseline (Table 2), 10-year ASCVD risk correlated

T A B L E 1 Baseline Characteristics

10-Year ASCVD Risk*

All <5% 5% to <7.5% 7.5% to <15% $15%


(N 5,805) (n 318) (n 521) (n 1,769) (n 3,197)

Age, yrs 68.9  6.0 61.8  2.8 63.6  4.0 66.2  4.7 71.9  5.1
Male 44 10 23 36 54
Diabetes 15 0 2 6 23
Current smokers 9 1 17 7 11
Hypertension 62 16 30 52 78
Systolic BP, mm Hg 139.5  18.5 120.9  12.6 127.4  13.5 134.7  16.0 145.9  18.1
Diastolic BP, mm Hg 78.2  9.1 74.0  8.2 75.8  7.9 78.1  8.9 79.1  9.2
Total cholesterol, mg/dl 202.5  38.6 204.6  36.0 205.6  35.7 206.8  37.3 199.4  39.7
HDL cholesterol, mg/dl 55.7  15.3 65.0  15.4 60.7  15.4 57.2  14.8 53.1  14.9
LDL cholesterol, mg/dl 114.2  33.2 112.5  30.2 114.6  31.1 117.7  32.7 112.3  34.0
Lipid-lowering medication 34 28 29 30 38
10-year ASCVD risk, % 19.3  12.3 3.8  0.8 6.3  5.0 11.1  2.2 27.4  10.8

Values are mean  SD or %. *Calculated by Pooled Cohort Equations.


ASCVD atherosclerotic cardiovascular disease; BP blood pressure; HDL high-density lipoprotein; LDL low-density lipoprotein.
JACC VOL. 68, NO. 9, 2016 Mortensen et al. 885
AUGUST 30, 2016:88191 Disease-Guided Initiation of Statin Therapy

T A B L E 2 Baseline Cardiovascular Risk Factors and Subclinical Atherosclerosis

CAC cPB

0 (n = 1,852) 199 (n = 1,675) $100 (n = 2,278) 0 (n = 1,315) 1299 (n = 2,212) $300 (n = 2,278)

Age, yrs 67.2  5.6 68.7  5.9 70.4  5.8 67.4  5.7 68.6  5.9 70.0  5.9
Male 29 41 58 34 48 55
Diabetes 11 14 18 13 13 17
Current smoker 6 8 10 4 7 12
Hypertension 55 60 69 56 58 69
Systolic BP, mm Hg 137.3  18.5 138.9  18.3 141.6  18.5 136.7  18.2 138.4  18.1 142.1  18.8
Diastolic BP, mm Hg 78.3  8.9 78.3  9.2 78.0  9.1 79.3  9.3 78.0  8.8 77.8  9.1
Total cholesterol, mg/dl 207.3  38.3 203.8  38.7 197.7  38.1 202.9  38.1 203.8  39.0 201  38.4
LDL cholesterol, mg/dl 117.6  32.5 114.7  33.5 110.9  33.3 114.1  32.5 114  33.2 114.0  33.6
HDL cholesterol, mg/dl 58.3  15.2 55.7  15.3 53.6  15.0 57.8  15.3 56.5  15.2 53.7  15.1
Lipid-lowering medication 27 35 40 45 34 38

Values are mean  SD or %, unless otherwise indicated.


CAC coronary artery calcium; cPB carotid plaque burden; other abbreviations as in Table 1.

directly with the amount of subclinical atheroscle- atherosclerosis, dened as CAC $100 or cPB $300,
rosis (Spearman correlation coefcient: 0.36 for CAC was similar in the coronary and carotid arteries (39%).
and 0.31 for cPB; p < 0.0001 for both). Absence of The risk-dependent NNS to identify 1 person
atherosclerosis was most common in lower-risk per- with absence or signicant atherosclerosis is shown
sons, whereas the opposite was true for the presence in Table 3. Regarding the potential for down-
of severe atherosclerosis (Figure 1). Furthermore, with classication, the NNS to nd 1 person with CAC 0
the methods used to assess subclinical disease in this was 2.6 and 4.5 among people with 10-year ASCVD
study, absence of atherosclerosis was more common risk of 7.5% to 15% and $15%, respectively. As for the
in the coronary arteries (32%) than in the carotid ar- potential for up-classication, the NNS to nd 1 per-
teries (23%). Among individuals with $7.5% 10-year son with CAC $100 among people with 10-year
ASCVD risk, 28% had no CAC and 20% had no ca- ASCVD risk of <5% and 5% to 7.5% was 10 and 5.3,
rotid plaque. The prevalence of signicant respectively. The same pattern was seen using

F I G U R E 1 Severity of Subclinical Atherosclerosis Stratied by 10-Year ASCVD Risk

100 CAC=0 100 cPB=0


CAC 1-99 cPB 1-299
CAC100 cPB300

80 80

64%
Prevalence (%)
Prevalence (%)

60 60
53%
50% 49%
48%
44% 42%
39% 42%
40 40
32% 32% 34%
31%30%
26% 28% 28%
26%
22% 20%
19%
20 20 17%
14%
10%

0 0
<5 5 to <7.5 7.5 to <15 15 <5 5 to <7.5 7.5 to <15 15
Pooled Cohort Equations Pooled Cohort Equations

The prevalence of coronary artery calcium (CAC) 0 declined with increasing risk calculated by the pooled cohort equations; the opposite happened for
CAC $100. A similar pattern was seen for carotid plaque burden (cPB). ASCVD atherosclerotic cardiovascular disease.
886 Mortensen et al. JACC VOL. 68, NO. 9, 2016

Disease-Guided Initiation of Statin Therapy AUGUST 30, 2016:88191

First, we assessed the consequences of down-


T A B L E 3 Risk-Dependent NNS for Subclinical Atherosclerosis
classifying those with a 10-year ASCVD risk of 7.5%
10-Year ASCVD Risk* to <15% from statin eligible to ineligible if CAC was 0,
All <5% 5% to <7.5% 7.5% to <15% $15% and up-classifying those with an intermediate
(N 5,805) (n 318) (n 521) (n 1,769) (n 3,197)
ASCVD risk of 5% to <7.5% if CAC was $100 from
CAC 0 32 64 53 39 22
optional to clear statin eligible (Table 5, Online
NNS 3.1 1.6 1.9 2.6 4.5
CAC $100 39 10 19 30 50 Tables 1 and 2). The specicity increased substan-
NNS 2.6 10 5.3 3.3 2.0 tially from 15% to 25% for both CHD and CVD without
cPB 0 23 44 32 26 17 any signicant loss in sensitivity, leading to a binary
NNS 4.3 2.3 3.1 3.8 5.9 NRI of 0.11 for CHD and 0.08 for CVD. Then we
cPB $300 39 14 20 32 49 expanded the candidate population for down-
NNS 2.6 7.1 5.0 3.1 2.0
classication to all with a 10-year ASCVD risk $7.5%
Values are % or NNS. *Calculated by Pooled Cohort Equations.
(not capped at 15%) and CAC 0. The gain in speci-
NNS number needed to screen to identify 1 individual with the value(s) in question; other abbreviations as in city doubled, resulting in a binary NRI of 0.20 for
Tables 1 and 2.
CHD and 0.14 for CVD (Table 5, Central Illustration).
These changes were driven mainly by down-
classifying the subpopulation with CAC 0 (Online
cPB 0 and $300 as cut-points for down- and
Tables 1 to 4).
up-classication, respectively (Table 3).
Assessment of the binary disease-guided reclassi-
CLINICAL EVENTS. Over a median follow-up of 2.7 cation approach using cPB cut-points of 0 and 300
years, 91 patients had a rst CHD event (MI, n 34; (equivalent to CAC 100) is shown in Table 5 and
UA, n 18; coronary revascularization without MI or Online Tables 5 to 8. Limiting down-classication
UA, n 39) and 138 patients had a rst CVD event (withholding statins) to those with 7.5% to <15% 10-
(in total: MI, n 34; UA, n 18; coronary revascu- year ASCVD risk decreased sensitivity to nearly the
larization, n 39; ischemic stroke, n 30; cardio- same extent as specicity increased, giving rise to no
vascular death, n 27). net effect on NRI. However, if everyone with a 10-
There was a strong relationship between subclini- year ASCVD risk $7.5% was down-classied to statin
cal atherosclerosis and clinical events (Table 4). Event ineligible if cPB was 0, the specicity increased more
rates were low in participants without subclinical than the sensitivity decreased (NRI 0.09 for CHD
atherosclerosis, even those with diabetes. Only 1 and and 0.06 for CVD) (Table 5, Central Illustration).
2 clinical events were seen among patients with dia- Excluding coronary revascularization from the
betes with CAC 0 and cPB 0, respectively. Kaplan- endpoints did not change the results. The HR for
Meier cumulative-event curves for CHD and CVD are CAC $100 and CHD events tended to be slightly lower
shown in Figure 2. (Online Table 9), but CAC- and cPB-guided NRIs
DISEASE-GUIDED RECLASSIFICATION. The ACC/AHA remained similar for both CHD and CVD events,
risk-based approach to statin allocation had high driven by a substantial increase in specicity with no
sensitivity (96%) but low specicity (15%) (Table 5). or minimal loss in sensitivity (Online Table 10).

T A B L E 4 Relationship Between Subclinical Atherosclerosis and Clinical Events

CHD Events CVD Events

Subclinical Total Event Rate per 1,000 HR* Event Rate per 1,000 HR*
Atherosclerosis N (%) n (%) Person-Years (95% CI) (95% CI) n (%) Person-Years (95% CI) (95% CI)

CAC
0 1,852 (32.0) 4 (0.2) 0.85 (0.322.26) 1 (reference) 15 (0.8) 3.19 (1.925.29) 1 (reference)
199 1,675 (29.0) 18 (1.1) 4.11 (2.596.52) 4.77 (1.3715.55) 25 (1.5) 5.71 (3.868.45) 1.48 (0.752.92)
$100 2,278 (39.0) 69 (3.0) 11.73 (9.2614.85) 13.73 (4.2344.59) 98 (4.3) 16.70 (13.7020.36) 3.98 (2.207.18)
cPB
0 1,315 (23.0) 6 (0.5) 1.74 (0.783.88) 1 (reference) 15 (1.1) 4.37 (2.637.25) 1 (reference)
1299 2,212 (38.0) 27 (1.2) 4.63 (3.166.81) 2.19 (0.895.39) 36 (1.6) 6.43 (4.638.91) 1.23 (0.672.26)
$300 2,278 (39.0) 59 (2.6) 9.94 (7.7112.84) 3.69 (1.558.75) 87 (3.8) 14.71 (11.9218.15) 2.14 (1.203.78)

*Adjusted for age, sex, total cholesterol, LDL cholesterol, HDL cholesterol, systolic BP, smoking, lipid-lowering medication, race, and diabetes. cPB 300 mm2 corresponds to
the percentile of CAC 100.
CHD coronary heart disease; CI condence interval; CVD cardiovascular disease; HR hazard ratio; other abbreviations as in Tables 1 and 2.
JACC VOL. 68, NO. 9, 2016 Mortensen et al. 887
AUGUST 30, 2016:88191 Disease-Guided Initiation of Statin Therapy

F I G U R E 2 Cumulative Incidence of Clinical Events

Cumulative CHD by Coronary Artery Calcium (CAC)


8 8 Cumulative CVD by Coronary Artery Calcium (CAC)
p-value <0.0001
p-value <0.0001
Cumulative Incidence (%)

Cumulative Incidence (%)


6 6

CAC 100 CAC 100


CAC 1-99 CAC 1-99
4 CAC = 0 4 CAC = 0

2 2

0 0
0 1 2 3 0 1 2 3
Years to Event Years to Event

Cumulative CHD by Carotid Plaque Burden (cPB) Cumulative CVD by Carotid Plaque Burden (cPB)
8 p-value <0.0001 8 p-value <0.0001
Cumulative Incidence (%)

Cumulative Incidence (%)

6 6
cPB 300 cPB 300
cPB 1-299 cPB 1-299
cPB = 0 cPB = 0
4 4

2 2

0 0
0 1 2 3 0 1 2 3
Years to Event Years to Event

In the Kaplan-Meier cumulative-event curves for coronary heart disease (CHD) and cardiovascular disease (CVD) events, CAC 0 and cPB 0 were
associated with low event rates. Abbreviations as in Figure 1.

T A B L E 5 Binary Disease-Guided Reclassication Approaches* to Statin Allocation

Sensitivity Specicity DSensitivity DSpecicity Reclassication


Predicted Outcome
and Risk Model % % % p Value % p Value NRI p Value

CHD (n 91)
ACC/AHA risk based 96 15 Reference Reference Reference
CAC guided (PCE risk <15%) 97 25 1 0.56 10 <0.0001 0.11 <0.0001
CAC guided (no upper risk limit) 94 37 2 0.42 22 <0.0001 0.20 <0.0001
cPB guided (PCE risk <15%) 91 21 4 0.04 6 <0.0001 0.02 0.42
cPB guided (no upper risk limit) 89 31 7 0.01 16 <0.0001 0.09 0.001
CVD (n 138)
ACC/AHA risk based 96 15 Reference Reference Reference
CAC guided (PCE risk <15%) 94 25 2 0.26 10 <0.0001 0.08 <0.0001
CAC guided (no upper risk limit) 88 37 8 0.005 22 <0.0001 0.14 <0.0001
cPB guided (PCE risk <15%) 91 21 4 0.01 6 <0.0001 0.02 0.30
cPB guided (no upper risk limit) 86 31 10 0.0002 16 <0.0001 0.06 0.04

*Cut-points for disease-guided reclassication are CAC 0 or cPB 0 for down-classication from statin eligible to ineligible and CAC $100 or cPB $300 mm2 for up-
classication from optional to clear statin eligible; 2 strategies are shown for each imaging modality, with and without an upper risk limit for down-classication.
ACC/AHA American College of Cardiology/American Heart Association; NRI net reclassication index (Dsensitivity Dspecicity); PCE Pooled Cohort Equations; other
abbreviations as in Tables 2 and 4.
888 Mortensen et al. JACC VOL. 68, NO. 9, 2016

Disease-Guided Initiation of Statin Therapy AUGUST 30, 2016:88191

SENSITIVITY ANALYSIS. In study participants who PREVIOUS VERSUS NEW ACC/AHA GUIDELINES. The
were not on lipid-lowering drugs at baseline exami- previous ACC/AHA guidelines for cardiovascular risk
nation (n 3,814), 60 CHD and 95 CVD events were assessment found it reasonable to measure CAC
observed during follow-up (Online Table 11). The and carotid intima-media thickness (cIMT) for risk
relationship between subclinical atherosclerosis and assessment in asymptomatic adults at intermediate
clinical events (HRs, Online Table 12) and disease- risk (14). This Class IIa recommendation for imaging-
guided reclassication (NRI up to 0.20 for CHD and guided risk assessment was, however, changed to
0.14 for CVD, Online Table 13) remained similar to the Class IIb for CAC and Class III for cIMT in the 2013 ACC/
overall results. AHA guidelines (2), primarily because the evidence for
Among individuals already on lipid-lowering drugs clinical utility vanished with the lower threshold for
at baseline examination, those with CAC 0 or statin therapy (dwarng the intermediate-risk group
cPB 0 had low event rates per 1,000 person-years [16]) and the change of predicted outcome from CHD to
(0 or 0 for CHD and 3.2 or 3.1 for CVD, respectively), ASCVD (CHD and stroke combined). The 2013 ACC/
and those with CAC $100 or cPB $300 mm 2 had AHA Class III recommendation for cIMT was based on
relatively high event rates (10.6 or 9.7 for CHD and evidence provided by a single meta-analysis of the
13.1 or 11.9 for CVD, respectively). incremental predictive value of the mean common
cIMT without separate plaque detection and assess-
DISCUSSION
ment, thus disregarding the known predictive power
of carotid plaque imaging (17).
In the contemporary BioImage cohort, 86% had a 10-
In the present study, rather than reviving a larger
year ASCVD risk $7.5% and qualied for primary
intermediate group population of uncertain risk, we
prevention with statins. This ACC/AHA risk-based
accepted the new Class I indication for risk-based
approach to statin allocation showed high sensi-
statin therapy ($7.5% 10-year ASCVD risk estimated
tivity (96%) but low specicity (15%). Event rates
by the ACC/AHA risk calculator) as a useful starting
were low in those without detectable atherosclerosis.
point, and then reclassied people across the 7.5% risk
Using a simple disease-guided reclassication
threshold guided by the absence or presence of sub-
approach based on well-dened cut-points for CAC
clinical atherosclerosis. Reclassication from statin
(and corresponding cPB cut-points) led to a substan-
eligible to ineligible and vice versa relied solely on
tial gain in specicity (less overtreatment) with no or
disease-based cut-points (CAC of 0 and 100) that have
only minor loss in sensitivity. In this elderly cohort,
documented incremental predictive value beyond
CAC performed better than cPB. The binary NRI was
traditional risk factor scoring (1821). This approach
up to 0.20 for CHD events, driven primarily by with-
performed well for both CHD and CVD outcomes,
holding statins in those with CAC 0. Limiting pri-
probably because most CVD events are CHD events.
mary prevention with statins to individuals with CAC
Individualized statin therapy in the elderly is
>0 could spare 1 in 4 elderly patients from taking
important for several reasons. Although universal
medication that will benet only a few.
treatment of all Americans ages 75 to 94 years might
STATIN THERAPY FOR ALL ELDERLY PEOPLE? Because be cost effective (22), elderly patients are more
ASCVD risk is strongly age-dependent, Wald and Law vulnerable to statin-related adverse effects because
(15) suggested in 2003 that everyone $55 years of age of comorbidity, polypharmacy, and risk for functional
should be offered treatment with safe risk-reducing limitation and possible cognitive impairment. Long-
drugs. By introducing a treatment threshold as low term effects of using low-dose computed tomogra-
as 7.5% 10-year ASCVD risk, the ACC/AHA guidelines phy for CAC screening are less concerning in the
indirectly supported such a universal treatment elderly. Overall, such considerations are obvious
principle because everyone with optimal risk factors topics for the patient-clinician discussion espoused
will now qualify for statin therapy if they live long by recent guidelines (2).
enough (African-American [AA] men age 66 years,
non-AA men age 63 years, AA women age 70 years, DOWNWARD RECLASSIFICATION IN THE ABSENCE
and non-AA women age 71 years) (3,5). However, the OF ATHEROSCLEROSIS. The Class IIb option in the
ACC/AHA guidelines remind clinicians that primary 2013 ACC/AHA guidelines for using CAC for rened
prevention with statins should not only depend on risk assessment when a risk-based treatment decision
predicted risk, but also include a discussion with the is uncertain applies only for up- and not down-
individual patient regarding potential treatment classication of risk. In fact, these guidelines pro-
benets, adverse effects, drugdrug interactions, and vide little guidance on how to personalize risk
patient preferences (2). assessment to avoid overtreating healthy individuals
JACC VOL. 68, NO. 9, 2016 Mortensen et al. 889
AUGUST 30, 2016:88191 Disease-Guided Initiation of Statin Therapy

who formally become statin eligible just because of risk <7.5%. In this lower-risk subpopulation, we
normal aging. observed 4 CHD and 6 CVD events during follow-up, 2
Although most cardiovascular events occur in old of which would have been identied by the CAC-based
age, many elderly patients do not have the underlying screening approach, none by the cPB-based screening
disease (atherosclerosis) responsible for the age- approach. Thus, the intermediate-risk strategy sup-
related risk for ASCVD. In the BioImage population ported by the previous ACC/AHA guidelines (14) may
(mean age 69 years), 86% qualied for risk-based still be relevant as assessed by the NNS, but relatively
statin therapy, but imaging revealed that up to one- few elderly will benet from such a strategy.
third had no CAC, which was associated with a low In the younger MESA cohort, the performance of
cardiovascular event rate, even in patients with dia- the guideline-recommended cut-points for up-
betes. Thus, there was a major gain in specicity (less classication was recently tested using recalibrated
overtreatment) by down-classifying those with $7.5% PCEs (25), which is applicable only to MESA and
10-year ASCVD risk estimated by PCE from statin therefore not generalizable to routine clinical practice
eligible to ineligible if CAC was 0, not only in those like the present disease-guided proposal. In the MESA
with a borderline increased risk (PCE risk from 7.5% to cohort, only a small subgroup (6.8%) was up-
15%), but also in everyone with a PCE risk $7.5%. Thus, classied, becoming eligible for primary prevention
our study conrmed that CAC 0 in asymptomatic with statins.
people without known ASCVD is associated with a very DISEASE-GUIDED RECLASSIFICATION. The clinical
low cardiovascular risk, called the power of zero (19). utility of detecting CAC and carotid plaque differs
Our data extended this observation to a contemporary (26). If the goal is to predict CHD events, CAC is likely
elderly population with a 10-year ASCVD risk well the optimal imaging test, whereas carotid plaque
above the 7.5% (Class I) risk threshold for statin assessment may be superior for stroke prediction (27).
therapy. However, PCE was devised to predict CHD and stroke
Few studies have evaluated the potential effect of combined (ASCVD), and this study is the rst to
atherosclerosis imaging on statin allocation after compare the incremental predictive value of CAC and
formal risk assessment as recommended by the new carotid plaque imaging beyond formal PCE-based risk
ACC/AHA guidelines. In the less contemporary MESA assessment. Judged by the NRI, CAC performed better
(Multi-Ethnic Study of Atherosclerosis) cohort of in- than cPB in terms of predicting both CHD events
dividuals who were on average 10 years younger than alone and the combined ASCVD outcome.
those in our study, of those eligible for statin therapy A major strength of the present analysis is that we
per the 2013 ACC/AHA guidelines, 41% had CAC 0 did not evaluate the clinical performance of imaging
and a low ASCVD event rate (5.2 events per 1,000 cut-points derived from the BioImage study, but
person-years of follow-up) (23). Among a small instead used CAC cut-points of proven value across
excluded subgroup of MESA participants 75 years of multiple populations (1821). We compared CAC and
age, almost all of whom were statin eligible, 18% had cPB with the same proportion of participants in the top
CAC 0 and a low 10-year event rate. In an even category dened by imaging. Although the cPB cut-
younger cohort from the Framingham Heart Study, a point of 300 mm 2 corresponding to CAC 100 may
CAC score of 0 identied a large low-risk group (33%) not be the optimal cut-point for the up-classication of
among the ACC/AHA statin-eligible participants (21). risk based on cPB, the main drivers of the positive NRI
Our results from a contemporary elderly population in this cohort were CAC 0 and cPB 0.
agreed with these recent observations and provided As previously described in greater detail, we used a
clinically important performance measures for statin novel, highly sensitive ultrasound-based sweep
allocation, such as sensitivity, specicity, and NRI. method to detect atherosclerosis in the carotid ar-
UPWARD RECLASSIFICATION IN THE PRESENCE OF teries (13). It may not necessarily be advantageous in
SIGNIFICANT ATHEROSCLEROSIS. The ACC/AHA elderly people, most of whom have at least early
guidelines state that CAC $300 or $75th percentile for atherosclerotic changes somewhere in the vascula-
age, sex, and ethnicity may be considered for up- ture. Thus, in the BioImage cohort, more people had
classication of risk in selected individuals (Class IIb cPB >0 (77%) than CAC >0 (68%). However, in a
recommendation) (2). Although this approach may be younger cohort, early detection may be prioritized
useful in some younger people (24), there is little room together with the use of a method not requiring ra-
for meaningful up-classication of risk in older people, diation. The ongoing PESA (Progression of Early
as most already qualify for risk-based statin therapy. In Subclinical Atherosclerosis) study, in which
the BioImage cohort, only 14% had a 10-year ASCVD multiple vascular beds are screened for subclinical
890 Mortensen et al. JACC VOL. 68, NO. 9, 2016

Disease-Guided Initiation of Statin Therapy AUGUST 30, 2016:88191

atherosclerosis, will clarify this issue (28). Regarding CONCLUSIONS


the cost of an atherosclerosis imaging test, most
people are ready to pay the cost of a test to avoid In individuals who qualied for ACC/AHA risk-based
taking a pill each day for the rest of their lives (29). statin therapy, limiting treatment to those with CAC
Another strength of the present study was that it >0 or cPB >0 led to a substantial gain in specicity
built on the existing ACC/AHA guidelines, including (less overtreatment) with no or only a minor loss
the online risk calculator. Furthermore, the tested in sensitivity. There was not much room for disease-
disease-guided reclassication approach is simple, is based expansion of statin eligibility in this elderly
easy to implement clinically, is less dependent on a cohort. A simple, individualized, disease-guided ap-
well-calibrated risk calculator, uses the strong evi- proach to statin allocation could potentially spare a
dence for low ASCVD risk universally across pop- signicant proportion of asymptomatic elderly people
ulations if CAC is 0, and focuses on the growing from lifelong treatment of questionable benet.
elderly, resource-consuming, at-risk population.
REPRINT REQUESTS AND CORRESPONDENCE: Dr.
STUDY LIMITATIONS. PCEs were designed to predict
the natural history of ASCVD in the absence of inter- Erling Falk, Department of Cardiology, Aarhus University

vention. A problem BioImage shared with other Hospital, Palle Juul-Jensens Boulevard, DK-8200 Aarhus,

contemporary cohorts was the lower-than-expected Denmark. E-mail: erling.falk@clin.au.dk.

event rates, which may be at least partly explained


by the healthy volunteer effect and modern common PERSPECTIVES
use of preventive therapies in people free of ASCVD
(810,30). We included coronary revascularization in
COMPETENCY IN PATIENT CARE AND
the composite endpoints but recognize that knowl-
PROCEDURAL SKILLS: In elderly people with no
edge of a high CAC score shared with the patient
clinical history of atherosclerosis, the absence of
and/or physicians may in itself lead to revasculariza-
coronary calcication and carotid plaque is associated
tion. However, excluding revascularization from the
with a low risk of cardiovascular events and might
composite endpoints did not change the study con-
reasonably lead to a decision to forego statin therapy.
clusions. The durability of our short-term results may
be questioned, but growing evidence indicates that
TRANSLATIONAL OUTLOOK: More work is
the warranty period for CAC 0 may be as long as
needed to prospectively evaluate the outcomes of
15 years (20,31), likely indicating lifelong low risk in
clinical decision strategies that individualize selection
elderly people. CAC 0 is well-dened and repro-
of preventive therapy based on a combination of
ducible, whereas cBP 0 is less well-dened and
population-based risk estimates and efcient incor-
depends on the protocol used to scan the carotid ar-
poration of noninvasive imaging tests.
teries; our use of a sensitive ultrasound protocol may
explain why cPB 0 was less common than CAC 0.

REFERENCES

1. Goff DC Jr., Lloyd-Jones DM, Bennett G, et al. 5. Breslow JL. Perspective on the 2013 American 10. Rana JS, Tabada GH, Solomon MD, et al. Ac-
2013 ACC/AHA guideline on the assessment of Heart Association/American College of Cardiology curacy of the atherosclerotic cardiovascular risk
cardiovascular risk. J Am Coll Cardiol 2014;63: guideline for the use of statins in primary equation in a large contemporary, multiethnic
293559. prevention of low-risk individuals. Circ Res 2014;114: population. J Am Coll Cardiol 2016;67:211830.
75860.
2. Stone NJ, Robinson JG, Lichtenstein AH, et al. 11. Muntendam P, McCall C, Sanz J, et al. The
2013 ACC/AHA guideline on the treatment of 6. Grundy SM. Cholesterol management in older BioImage Study: novel approaches to risk assess-
blood cholesterol to reduce atherosclerotic car- persons. Am J Med 2014;127:11401. ment in the primary prevention of atherosclerotic
diovascular risk in adults. J Am Coll Cardiol 2014; 7. Navar-Boggan AM, Peterson ED, DAgostino RB, cardiovascular diseasestudy design and objec-
63:2889934. et al. Using age- and sex-specic risk thresholds to tives. Am Heart J 2010;160:4957.

3. American College of Cardiology, American Heart guide statin therapy: one size may not t all. J Am 12. Baber U, Mehran R, Sartori S, et al. Prevalence,
Association. ACC/AHA ASCVD risk calculator. Coll Cardiol 2015;65:16339. impact, and predictive value of detecting sub-
Available at: http://tools.acc.org/ASCVD-Risk- 8. Goff DC Jr., DAgostino RB, Pencina M, et al. clinical coronary and carotid atherosclerosis in
Estimator/. Accessed June 18, 2016. Calibration and discrimination among multiple asymptomatic adults: The BioImage study. J Am
cardiovascular risk scores in a modern multiethnic Coll Cardiol 2015;65:106574.
4. Pandya A, Sy S, Cho S, et al. Cost-
effectiveness of 10-year risk thresholds for cohort. Ann Intern Med 2015;163:68. 13. Sillesen H, Muntendam P, Adourian A, et al.
initiation of statin therapy for primary preven- 9. Pursnani A, Massaro JM, Hoffmann U. Predicted Carotid plaque burden as a measure of subclinical
tion of cardiovascular disease. JAMA 2015;314: vs observed clinical event risk for cardiovascular atherosclerosis: comparison with other tests for
14250. diseasereply. JAMA 2015;314:20823. subclinical arterial disease in the High Risk Plaque
JACC VOL. 68, NO. 9, 2016 Mortensen et al. 891
AUGUST 30, 2016:88191 Disease-Guided Initiation of Statin Therapy

BioImage study. J Am Coll Cardiol Img 2012;5: 21. Pursnani A, Massaro JM, DAgostino RB, et al. 27. Blaha MJ. The future of CV risk prediction:
6819. Guideline-based statin eligibility, coronary artery multisite imaging to predict multiple outcomes.
calcication, and cardiovascular events. JAMA J Am Coll Cardiol Img 2014;7:10546.
14. Greenland P, Alpert JS, Beller GA, et al. 2010
2015;314:13441.
ACCF/AHA guideline for assessment of cardiovas- 28. Fernndez-Friera L, Pealvo JL, Fernndez-
cular risk in asymptomatic adults. J Am Coll Car- 22. Odden MC, Pletcher MJ, Coxson PG, et al. Ortiz A, et al. Prevalence, vascular distribution,
diol 2010;56:e50103. Cost-effectiveness and population impact of sta- and multiterritorial extent of subclinical athero-
tins for primary prevention in adults aged 75 years sclerosis in a middle-aged cohort: the PESA (Pro-
15. Wald NJ, Law MR. A strategy to reduce car-
or older in the United States. Ann Intern Med gression of Early Subclinical Atherosclerosis)
diovascular disease by more than 80%. BMJ
2015;162:53341. study. Circulation 2015;131:210413.
2003;326:1419.
23. Nasir K, Bittencourt MS, Blaha MJ, et al. Im- 29. Hutchins R, Viera AJ, Sheridan SL, et al.
16. Blaha MJ, Blumenthal RS. New risk- Quantifying the utility of taking pills for cardio-
plications of coronary artery calcium testing
assessment guidelinesmore or less personal- vascular prevention. Circ Cardiovasc Qual Out-
among statin candidates according to American
ized? Nat Rev Cardiol 2014;11:1367. comes 2015;8:15563.
College of Cardiology/American Heart Association
17. Stein JH, Tattersall MC. Carotid intima-media cholesterol management guidelines: MESA (Multi- 30. Leening MJ, Heeringa J, Deckers JW, et al.
thickness and cardiovascular disease risk predic- Ethnic Study of Atherosclerosis). J Am Coll Cardiol Healthy volunteer effect and cardiovascular risk.
tion. J Am Coll Cardiol 2014;63:23012. 2015;66:165768. Epidemiology 2014;25:4701.
18. Hecht HS. A zero coronary artery calcium 24. Tota-Maharaj R, Blaha MJ, McEvoy JW, et al. 31. Valenti V, Hartaigh B, Heo R, et al. A 15-year
score: priceless. J Am Coll Cardiol 2010;55: Coronary artery calcium for the prediction of warranty period for asymptomatic individuals
111820. mortality in young adults <45 years old and without coronary artery calcium: a prospective
elderly adults >75 years old. Eur Heart J 2012;33: follow-up of 9,715 individuals. J Am Coll Cardiol
19. Blaha MJ, Silverman MG, Budoff MJ. Is there a
295562. Img 2015;8:9009.
role for coronary artery calcium scoring for man-
agement of asymptomatic patients at risk for 25. Yeboah J, Polonsky TS, Young R, et al. Utility
coronary artery disease? Clinical risk scores are not of nontraditional risk markers in individuals ineli-
KEY WORDS cardiovascular disease,
sufcient to dene primary prevention treatment gible for statin therapy according to the 2013
carotid plaque, coronary calcium, primary
strategies among asymptomatic patients. Circ American College of Cardiology/American Heart
prevention, risk assessment
Cardiovasc Imaging 2014;7:398408. Association cholesterol guidelines. Circulation
2015;132:91622.
20. Greenland P. When nothing is really some-
thing: new evidence of the importance of zero 26. Shah PK. Can carotid plaque predict A PP END IX For an expanded Methods section
coronary calcium. J Am Coll Cardiol Img 2015;8: coronary plaque? J Am Coll Cardiol Img 2013;6: and supplemental tables, please see the online
9102. 116871. version of this article.