Lipoprotein-associated phospholipase A₂ and future risk of subclinical disease and cardiovascular events in individuals with type 2 diabetes: the Cardiovascular Health Study

Abstract

Aims/hypothesis

Type 2 diabetes is an established risk factor for cardiovascular disease (CVD). This increased risk may be due in part to the increased levels of inflammatory factors associated with diabetes. Lipoprotein-associated phospholipase A₂ (Lp-PLA₂) is a risk marker for CVD and has pro-inflammatory effects in atherosclerotic plaques. We therefore sought to determine whether Lp-PLA₂ levels partially explain the greater prevalence of subclinical CVD and greater incidence of CVD outcomes associated with type 2 diabetes in the Cardiovascular Health Study.

Methods

We conducted a cross-sectional and prospective study of 4,062 men and women without previous CVD from the Cardiovascular Health Study (1989 to 2007). Lp-PLA₂ mass and activity were measured in baseline plasma. Subclinical disease was determined at baseline and incident CVD was ascertained annually. We used logistic regression for cross-sectional analyses and Cox proportional hazards models for incident analyses.

Results

At baseline, Lp-PLA₂ mass did not differ significantly by type 2 diabetes status; however, Lp-PLA₂ activity was significantly higher among type 2 diabetic individuals. Baseline subclinical disease was significantly associated with baseline diabetes and this association was similar in models unadjusted or adjusted for Lp-PLA₂ (OR 1.68 [95% CI 1.31–2.15] vs OR 1.67 [95% CI 1.30–2.13]). Baseline type 2 diabetes was also significantly associated with incident CVD events, including fatal CHD, fatal myocardial infarction (MI) and non-fatal MI in multivariable analyses. There were no differences in these estimates after further adjustment for Lp-PLA₂ activity.

Conclusions/interpretation

In this older cohort, differences in Lp-PLA₂ activity did not explain any of the excess risk for subclinical disease or CVD outcomes related to diabetes.

Introduction

Type 2 diabetes is an established risk factor for cardiovascular disease (CVD). This increased risk may be due in part to the increased levels of inflammatory factors associated with type 2 diabetes. Recently, there has been great interest in the macrophage-derived enzyme lipoprotein-associated phospholipase A₂ (Lp-PLA₂), as it may play a key role in atherosclerotic plaque inflammation [1]. Lp-PLA₂ is also an independent predictor of coronary heart disease and ischaemic stroke in the general population and in clinical populations, including patients with diabetes [2–4]. Although Lp-PLA₂ was originally known as platelet-activating factor (PAF) acetylhydrolase because it degrades PAF and therefore potentially protects against atherosclerosis [5], overwhelming evidence now suggests that this is outweighed by its proatherogenic associations [1].

Previously, Barzilay et al. had shown that glucose disorders are associated with an increased prevalence of total CVD and an increased proportion of clinical diabetes relative to subclinical disease in the Cardiovascular Health Study (CHS) cohort [6]. In addition, Kuller et al. reported that the risk of clinical cardiovascular events was greatest for participants with a history of diabetes compared with those with newly diagnosed diabetes at baseline in the CHS [7]. We therefore hypothesised that Lp-PLA₂ mass and activity would partially explain the greater prevalence of subclinical disease at baseline and the greater incidence of CVD outcomes associated with type 2 diabetes in the CHS cohort.

Methods

Study population

Participants were members of the CHS, a population-based cohort study of risk factors for CVD in older adults. The design and recruitment have been described in detail in previous publications [8, 9]. Overall, 5,888 adults aged 65 years and older were randomly sampled from Medicare eligibility lists in four US communities. All participants gave informed consent for participation in the study, and all procedures were conducted under institutionally approved protocols at each centre. Mean age at enrolment was 73 years (range 65–100), with 58% and 16% of participants being women and black, respectively. We excluded 1,517 participants who had CVD at baseline. We also excluded 64 participants for whom diabetes status was missing, and 239 and six participants without Lp-PLA₂ mass and Lp-PLA₂ activity data, respectively. This left 4,062 participants for the present study.

Self-reports of CVD outcomes were validated according to review of medical records [10]. Medication use was confirmed using an inventory at the home interview. Events were ascertained through 30 June 2007. Participants were categorised as having type 2 diabetes mellitus at baseline if they reported use of insulin or oral hypoglycaemic medications, or had a fasting glucose level of ≥7 mmol/l (126 mg/dl) or a random glucose measurement of ≥11 mmol/l (200 mg/dl). Participants who did not have prevalent CVD at baseline were categorised as having subclinical disease if they had any of the conditions listed for subclinical disease in Table 1 (footnotes). Prevalent CVD was defined as having previously had: myocardial infarction (MI), stroke, congestive heart failure (CHF), angina, percutaneous transluminal coronary angioplasty (PTCA), coronary artery bypass graft (CABG), claudication or transient ischaemic attack. Cardiovascular outcomes included fatal CHD, fatal or non-fatal MI, CHF, angina, CABG or PTCA.

Table 1.

Mean levels or frequencies of CVD risk factors by type 2 diabetes status at the 1989–1990 and 1992–1993 baseline visits

Variable	Diabetes	Non-diabetic	p value
n	541	3,521
Lp-PLA2 mass (ng/ml)	337.8±116.3	341.5±117.5	0.49
Lp-PLA2 activity (nmol min−1 ml−1)	40.1±12.6	38.6±12.7	<0.05
Maximum stenosis (≥25%), n (%)	263 (48.9)	1,438 (41.1)	<0.01
Subclinical disease, n (%)a	363 (68.8)	1,819 (52.7)	<0.001
BMI (kg/m2)	28.8±4.8	26.3±4.6	<0.0001
HDL-cholesterol (mmol/l)	1.2±0.3	1.5±0.4	<0.0001
LDL-cholesterol (mmol/l)	3.3±1.0	3.4±0.9	0.0910
Triacylglycerol (mmol/l)	1.6 (1.2, 2.3)	1.3 (1.0, 1.7)	<0.001
C-reactive protein (mg/l)	2.5 (1.4, 4.9)	1.7 (0.9, 3.0)	<0.0001

Unless otherwise indicated, values are mean±SD or median (25th percentile, 75th percentile)

p values represent difference in means or proportions between diabetic and non-diabetic participants

^aSubclinical was defined as not having clinical disease (apart from presence of ECG abnormalities), but having any of the following: ankle–arm index <0.9, internal carotid artery wall thickness >80th percentile, carotid stenosis >25%, major ECG abnormalities (based on Minnesota code) or a Rose questionnaire positive for claudication or angina pectoris in the absence of clinical diagnosis of angina pectoris and claudication [11]

GlaxoSmithKline (Research Triangle Park, NC, USA) measured plasma concentrations of Lp-PLA₂ activity using a tritium-labelled form of platelet-activating factor as substrate in a 96-well microplate, as described in detail previously [3]. Samples were tested in duplicate and were retested if the replicate coefficient of variation was greater than 25%. The analytical coefficient of variation was 7.5%. Plasma Lp-PLA₂ mass was measured at the CHS Central Laboratory at the University of Vermont using an enzymelinked immunosorbent assay kit (second-generation PLAC Test; diaDexus, South San Francisco, CA, USA). All samples were analysed in duplicate and the analytical coefficient of variation was 6.3%.

Statistical analysis

Univariate differences between diabetic and non-diabetic individuals at baseline were evaluated by χ² test for categorical covariates, by t tests for normal continuous covariates and by Mann–Whitney non-parametric tests for non-normal continuous covariates. Logistic regression was used to evaluate the cross-sectional association between subclinical disease and type 2 diabetes status. Cox proportional hazards models were used to estimate the relative risk (HR) of CVD outcomes predicted by type 2 diabetes status. Entry time into the analysis corresponded to the participants’ study enrolment date, with time-at-risk lasting until the earliest of incident CVD event, death, loss to follow up or the last day of adjudicated follow-up (30 June 2007). Using Cox regression, we also estimated the relative risk of CVD outcomes predicted by Lp-PLA₂ activity and mass, and stratified above or below the 75th percentile by type 2 diabetes status. All multivariable models were evaluated for multiple variables. In general, covariates were retained in the final models if they influenced the risk estimate by 10% or more. We evaluated potential mediation of these associations by Lp-PLA₂ activity by comparing models with and without adjustment for this factor. Analyses were performed using Stata 10.1 (Stata, College Station, TX, USA). All p values were two-tailed (α=0.05)

Results

Of the 4,062 participants, 37.4% were men and 84.3% were white. The mean age was 72.3±5.3 years. Table 1 shows the baseline characteristics of CVD risk factors by type 2 diabetes status. Participants with type 2 diabetes had more risk factors, including more subclinical disease, higher Lp-PLA₂ activity, lower HDL-cholesterol, higher triacylglycerol and higher C-reactive protein levels.

Baseline subclinical disease was significantly associated with baseline type 2 diabetes status in multivariable models (OR 1.68 [95% CI 1.31–2.15]). These associations were not significantly altered after adjusting for Lp-PLA₂ activity (OR 1.67 [95% CI 1.30–2.13]). When we stratified by Lp-PLA₂ above or below the 75th percentile, the associations were modestly but not significantly attenuated in those above the 75th percentile for mass and activity (data not shown).

During an average follow up of 12.8 (±4.9) years, type 2 diabetes, in multivariable models, was significantly associated with incident CVD outcomes including fatal CHD and MI, as well as non-fatal MI, CHF, angina and CABG. However, these associations were not altered after adjusting for Lp-PLA₂ activity (Table 2). Moreover, after stratifying by Lp-PLA₂ above or below the 75th percentile, the risk estimates between type 2 diabetes and incident CVD were attenuated in those above the 75th percentile, albeit not significantly (data not shown). Finally, we considered the association between Lp-PLA₂ and incident CVD, stratified by type 2 diabetes status. Similarly to other studies, increased risk of CVD outcomes was associated with higher Lp-PLA₂ activity and mass, but these risks were not modified by type 2 diabetes status (Electronic supplementary material [ESM] Table 1).

Table 2.

Cox proportional hazards analysis for the association between incident CVD outcomes and type 2 diabetes status

Variables per outcome	HRs not adjusted for Lp-PLA2 activity		HRs adjusted for Lp-PLA2 activity
	No diabetes	Diabetes	No diabetes	Diabetes
Fatal CHD
Events (n)	378	118
Person-years	45,594	5,704
Adjusted HRa	1.0	2.55 (2.07–3.14)	1.0	2.54 (2.06–3.13)
Multivariable HRb	1.0	2.14 (1.69–2.70)	1.0	2.14 (1.69–2.70)
Fatal MI
Events (n)	51	15
Person-years	46,203	5,955
Adjusted HRa	1.0	2.51 (1.40–4.50)	1.0	2.51 (1.40–4.49)
Multivariable HRb	1.0	2.10 (1.12–3.94)	1.0	2.10 (1.12–3.94)
Non-fatal MI
Events (n)	414	95
Person-years	44,251	5,575
Adjusted HRa	1.0	1.87 (1.49–2.34)	1.0	1.85 (1.48–2.32)
Multivariable HRb	1.0	1.56 (1.21–2.01)	1.0	1.56 (1.21–2.01)
CHF
Events (n)	826	195
Person-years	43,359	5,194
Adjusted HRa	1.0	2.07 (1.77–2.43)	1.0	2.06 (1.76–2.42)
Multivariable HRb	1.0	1.52 (1.28–1.82)	1.0	1.52 (1.28–1.82)
CVD outcomec
Events (n)	1,239	265
Person-years	40,019	4,633
Adjusted HRa	1.0	1.88 (1.64–2.15)	1.0	1.87 (1.64–2.14)
Multivariable HRb	1.0	1.53 (1.32–1.77)	1.0	1.53 (1.32–1.77)

^aHR (95% CI), adjusted for age, sex and race

^bHR (95% CI), adjusted for age, sex, race (black, non-black), BMI, smoking status (former, never, current), smoking history (lifetime pack-years), alcohol use (never, former, <7 drinks/week, >7 drinks/week), HDL-cholesterol, LDL-cholesterol, use of statins, C-reactive protein, field centre, education (lower than high school, high school or General Education Diploma, higher than high school), physical activity (energy expenditure), creatinine and hypertension

^cAny of the above (i.e. fatal CHD, fatal MI, non-fatal MI, CHF)

Discussion

In this study, Lp-PLA₂ mass or activity did not account for any part of the greater prevalence of subclinical disease or the greater incidence of CVD outcomes associated with type 2 diabetes in the CHS cohort of older adults. As shown previously in this cohort, Lp-PLA₂ mass and activity predicted CVD events [2]; however, these associations did not differ by type 2 diabetes status, as shown in the present study.

Importantly, a recently published study by Hatoum et al. found that type 2 diabetic participants with elevated levels of Lp-PLA₂ activity were significantly more likely to develop CHD than those without elevated levels [4]. This study included participants in the Health Professionals Follow-up Study and the Nurses’ Health Study. Its finding is counter to what we found here, where type 2 diabetic participants in the upper 75th percentile of Lp-PLA₂ activity were not more likely to develop any CVD outcome. Perhaps the differences in age between these cohorts (i.e. ~73 years vs ~60 years in the study of Hatoum et al.) explains the difference in findings. An important limitation of the other study [4] is that the investigators did not determine whether the association between Lp-PLA₂ and CHD in diabetic participants was different from that in non-diabetic participants. We, in the present study, did not find that Lp-PLA₂ altered the association between type 2 diabetes and incident CHD outcomes when compared with non-diabetic participants (see Table 2).

It is possible that Lp-PLA₂ is not sufficiently sensitive as a single marker, given all the other factors that contribute to CVD risk in older diabetic individuals. Other limitations include the fact that Lp-PLA₂ and type 2 diabetes status were measured at the same time; a later measurement of Lp-PLA₂ might, conceivably, have resulted in a stronger association with CVD outcomes. Future studies need to address the temporality of these relationships, as it was not possible to determine whether higher Lp-PLA₂ levels preceded or followed the development of type 2 diabetes in this study, which was not designed to test causality. Furthermore, because we studied an older population with type 2 diabetes, those who died earlier in life from diabetes-related complications could not be studied, but may have had higher Lp-PLA₂ levels. It is also possible that measurement error in the assays for Lp-PLA₂ could have influenced these null findings or that the diabetic participants in this study had adequate amounts of anti-oxidants to counter elevated levels of oxidised LDL. Despite these limitations, this study does have several strengths, including a large sample size and number of incident CVD events, with 100% follow-up and measurement of numerous confounding factors.

Overall, our results suggest that Lp-PLA₂ does not mediate the association between type 2 diabetes and subclinical CVD and CVD events. Also, type 2 diabetes status does not modify the association between Lp-PLA₂ and CVD outcomes in an older population.

Abbreviations

CABG

Coronary artery bypass graft

CHF

Congestive heart failure

CHS

Cardiovascular Health Study

CVD

Cardiovascular disease

Lp-PLA₂

Lipoprotein-associated phospholipase A₂

MI

Myocardial infarction

PTCA

Percutaneous transluminal coronary angioplasty