Cholesterol efflux capacity does not associate with coronary calcium, plaque vulnerability and telomere length in healthy octogenarians

: 185 words Abstract Several studies revealed that traditional risk factors are less effective in predicting CVD risk in the elderly, suggesting the need to identify new biomarkers. Here we evaluated the association between serum cholesterol efflux capacity (CEC), an atheroprotective property of HDL recently identified as a novel marker of CVD risk, and atherosclerotic burden in a cohort of very old, healthy individuals. Serum CEC values were not significantly correlated neither with calcium score nor with markers of vulnerable plaque, such as positive remodeling, hypodensity, spotty calcification or napking-ring sign. In addition, no association was detected between CEC and telomere length, a marker of biological aging that has been linked to atherosclerosis extent. Interestingly, elderly subjects presented a remarkably higher CEC (+30.2%; p<0.0001) compared to values obtained from a cohort of sex-matched, free of cardiovascular events, middle-aged individuals. In conclusion, serum CEC is not related to traditional risk factors in very old, free of cardiovascular events subjects, but has significantly higher values compared to a healthy, younger population. Whether this improved HDL functionality may represent a protective factor in CVD onset has to be established in future studies.


Introduction
In the last decades, the worldwide population exhibited an increasing life expectancy with a consequent rise in the elderly population (1). Aging is accompanied by a progressive deterioration of the physiological functions in several organs, including the cardiovascular apparatus (2). In particular, CVD is the leading cause of morbidity and mortality in individuals over the age of 65 and aging itself is one of the most powerful cardiovascular risk factors, as indicated by the scores published by recent guidelines (3).
Interestingly, traditional lipid risk factors might be less effective in predicting CVD risk in the elderly compared to middle-aged. For instance, the Framingham Risk Score, that estimates the individual cardiovascular disease risk at10-years and takes into account the contributions of all serum lipoproteins and additional risk factors, has shown to have only a mild ability to identify CVD risk among old people with longevity potential (4). Focusing on plasma lipids, the positive association between total cholesterol (TC) and cardiovascular risk is strong in middle-aged adults, but it becomes weaker in those with 80 years of age or older (5). In addition, we recently reported that high LDL cholesterol (LDL-C) levels are not associated to subclinical coronary artery disease in healthy octogenarians (6) and results of a recent meta-analysis showed either a lack or an inverse association between LDL-C and both all-cause and CVD mortality in the late-life age (7).
Whether HDL cholesterol (HDL-C) level has a predictive power of CVD in old people is still a matter of debate.
Some findings suggest that plasma levels of these lipoproteins may be a more valuable predictor of cardiovascular events compared to TC, LDL-C and non HDL-C level in older age (8). This concept is supported by studies that observed an inverse association between HDL-C levels and the risk of coronary artery disease or ischemic stroke in the elderly (9,10), including a recent work in which we demonstrated that low HDL-C is independently associated with subclinical coronary atherosclerosis in a cohort of healthy octogenarians (6).
Consistently, a prospective study showed that persistently low HDL-C levels were a risk factor for the development of CVD events in the elderly (11). On the contrary, other data reported weaker or lack of association between HDL-C and CVD risk in old people (12, 13). Overall, this decline in the predictivity of the traditional risk factors with aging makes the stratification in elderly more difficult compared to younger individuals, suggesting an urgent need to identify new biomarkers able to provide a more precise estimate of cardiovascular risk in these subjects.
Among the novel and recently proposed cardiovascular risk biomarkers there is the HDL cholesterol efflux capacity (CEC), the most established HDL antiatherogenic property (14,15), consisting in the capacity to promote the release of excess cholesterol from macrophages of the arterial wall (16). A growing number of evidence indicates an inverse relationship between HDL CEC and both the prevalence and the incidence of CVD (17)(18)(19). Interestingly, this association has been found to persist even after adjustment for traditional risk factors including HDL-C plasma levels, suggesting that HDL function may represent a predictor of CVD transcending the risk estimation based solely on the amount of particles present in the bloodstream. In support of this concept, by guest, on July 19, 2018 www.jlr.org Downloaded from admission into the study and stored at -80°C until use. Before the addition to cells, sera were gently thawed in ice. After the efflux period, the medium was removed from each well, filtered to remove cellular debris, and 3 H cholesterol quantified by liquid scintillation counting. Percentage efflux was determined as the radioactivity detected in the media divided by the total radioactivity incorporated by cells. All samples were run in triplicate.
To check for adequate cAMP-induced ABCA1 expression, the specific ABCA1 cholesterol acceptor lipid-free human apoA-I was tested together with serum samples. Values were normalized by dividing the efflux capacity of individual subjects by the efflux capacity of a standard serum run in each assay (19). The direct comparison between octogenarians and middle-aged subjects in terms of CEC was feasible because the same serum standard was used for normalization in all the assays.

Cardiac computed tomography
Cardiac computed tomography was performed to evaluate Coronary Calcium Score on a 64-detector row scanner (Aquilion 64, Toshiba, Ottawara, Tokyo, Japan). Axial slices of 3-mm thickness were acquired in synchrony with an electrocardiographic tracing in 70% of the RR interval. Coronary calcifications were defined by at least three continuous pixels with a minimum of 130 Hounsfield units, and were analyzed by a single certified radiologist. The Agatston method was used to express the values of coronary calcification. Prior to the scans, all patients with heart frequency above 60 beats per minute received a maximum dose of 25 mg endovenous cardioselective beta-blocker metoprolol (Seloken ® -Astra-Zeneca) and 5 mg Isosorbide dinitrate (Isordil ® -EMS). For Coronary Computed Tomography Angiography analyses, the coronary arteries were grouped in four major arteries named: left main trunk, left anterior descendent, left circumflex and right coronary artery, including their major and minor branches. Coronary atherosclerotic plaques were defined as any tissue >1mm² within or adjacent to the lumen that could be discriminated from surrounding pericardial tissue, epicardial fat or lumen and that could be identified in at least 2 planes. Plaque morphology was visually described as 1) noncalcified plaque, if there is no signal of higher density in the plaque; 2) calcified plaque, if calcification larger than 3 mm was detected; 3) spotty calcified plaques, if calcification present <3 mm in size on curved multiplane reformatting images and occupied only 1 side on sectional images (28) and 4) Positive Remodeling plaques, when the diameter in the vessel diameter at the plaque site was at least 10% larger than the segment reference set proximal to the lesion in a normal vessel segment. The presence or not of vulnerable features in the coronary plaques was evaluated by using the Motoyama criteria (28). For each coronary artery (left main, right coronary, circumflex coronary, anterior descending coronary) one point was given if they have the characteristic. Hence, plaque remodeling was counted from 0 (absence) to 4 (presence in the four coronary branches). The same was made for hypodensity and presence of calcium spots. Napkin-ring sign was identified by the presence of a ring of high attenuation around the coronary artery plaque and ring attenuation presenting greater than those of the adjacent plaque and not greater than 130 Hounsfield units (29).

Telomere length
Telomere length (TL) was estimated according to a previously described method (30). Briefly, genomic DNA was extracted from leukocyte homogenate using Qiagen DNA extraction kit and telomere length determined by real-time RT-PCR. Amplification reaction was composed by 35 ng of DNA Sybr Mastermix (Invitrogen) and primers for telomere (forward 5'CGGTTTGTTTGGGTTTGGGTTTGGGTTTGGG TTTGGGTT3 'and backward 5'GGCTTGCCTTACCCTTACCCTTACCC TTACCCTTACCCT3') and for housekeeping control, the 36b4, (forward 5'CAGCAAGTGGGAAGGTGTAATCC3 'and backward 5'CCCATTCTATCATCAACGGGTACAA3'). Fluorescence detection was performed with Rotor-Gene 6000 (Corbett Research). For each DNA sample, we performed 3 quantitative PCR for telomere repeat sequence (T) and 3 for the reference single copy gene (S), i.e. the 36b4 gene. The TL was estimated by the average of the 3 T measurements divided by the average of the 3 S measurements. TL values were available for 51/59 subjects.

Statistical analysis
Data distribution was assessed for normality with the use of both histograms and Shapiro Wilk test. Normally distributed data are presented as mean ± standard deviation and non-normally distributed data are presented in the form of median (interquartile range). Gaussian curves were created by nonlinear regression of the frequency distribution. Proportional differences between groups were evaluated using chi-square. Mean differences between groups were evaluated with one-way ANOVA and analysis of covariance (ANCOVA) where there was need for adjustments. Non-parametric data were compared by using Mann-Whitney unpaired t-test. In order to keep the comparison between middle-aged adults and elderly balanced and unsaturated, a propensity score was generated using age, sex and variables that showed significant or marginal differences in the unadjusted comparison between these two groups: mean blood pressure, presence of diabetes, apoB, apoA-I, creatinine, triglycerides and glucose. Hence, CEC values were compared between these two groups by ANCOVA adjusting for the propensity score. Statistical analyses were conducted using SPSS software version 21.0 or Graph Pad Prism software (version 6.01).

RESULTS
The baseline demographic and clinical characteristics of elderly participants are summarized in Table 1. In agreement with the overall demographics in this age group, most octogenarians in the study were women. To rule out comorbidities that could indirectly interfere with markers of atherosclerotic risk, such as cancer or inflammatory diseases, we used strict selection criteria. Thus, beyond the absence of these diseases, lipid profile, glucose, uric acid, fibrinogen, CRP and creatinine were within normal reference ranges. Calcium supplement plus bisphosphonate was used by eleven elderlies, whose individual characteristics are reported in Supplemental   Table S1. Calcitriol was not used by enrolled patients. The use of calcium plus bisphosphonate was not associated with CAC (p = 0.48). Over 30% of the population had CAC=0. The coronary calcium score distribution in male and female subjects is shown in Supplemental Figure S1. Table 1 also shows the presence or absence of remodeling, hypodensity, spotty calcification and napkin-ring sign, markers of unstable plaques in the arteries of analyzed subjects.
In order to explore the potential relationship between serum CEC and CAC a in the elderly subjects, we stratified CEC values between subjects with CAC = 0 and with CAC > 0 (31). As it can be seen in Figure 1, no significant difference was observed between the two groups (average CEC ratio over standard serum was 1.47 ± 0.15 for CAC = 0, compared to 1.40 ± 0.16 for CAC > 0; p = 0.09). We further stratified CEC values according to tertiles of CAC. We observed that CEC values at the first CAC tertile were higher compared to the second (p = 0.017), but not to the third tertile (Supplemental Figure S2).
We subsequently evaluated the relationship between serum CEC and the presence of atherosclerotic plaques with positive remodeling, hypodensity, spotty calcification or napkin-ring sign. Individuals were categorized depending on the absence or presence of vulnerable plaques markers and their serum CEC values were compared. As shown in Figure 2, comparable CEC values were displayed by subjects negative or positive for arterial remodeling (average CEC ratio over standard serum was 1.41 ± 0.14 compared to 1.45 ± 0.18, ns, Figure   2A), hypodensity (average CEC ratio over standard serum was 1.42 ± 0.15 compared to 1.43 ± 0.17, ns, Figure   2B) calcium spot (average CEC ratio over standard serum was 1.41 ± 0.16 compared to 1.45 ± 0.16, ns, Figure   2C) and napkin-ring sign (average CEC ratio over standard serum was 1.42 ± 0.14 compared to 1.45 ± 0.25, ns, Figure 2D).
We also evaluated whether serum CEC is associated with TL in our population, considering that telomere shortening, occurring with aging, has been previously related to early markers of cardiovascular disease and subclinical atherosclerosis (32,33). Similar to what observed for the parameters considered above, individuals with TL values under and over the median were not different in terms of serum CEC: average CEC ratio over standard serum was 1.44 ± 0.16 and 1.40 ± 0.18 in subjects with TL values under and over the median, respectively (ns, Figure 3). No differences were detected by further stratifying CEC values by tertiles of TL (ANOVA p-value = 0.968, Supplemental Figure S3). In addition, TL did not associate neither with the measured markers of subclinical atherosclerosis (flow-mediated dilation and intima-media thickness) nor with CAC (data not shown).
We finally compared serum CEC values of elderly subjects with those obtained from a cohort of sex-matched, free of cardiovascular events, middle-aged individuals (n = 140) that we analyzed in a previous study (18), whose demographic and clinical characteristics are reported in Table 2.
As expected according to the established age-dependent variations of blood pressure (34), middle-aged individuals presented lower values of systolic blood pressure and higher values of diastolic blood pressure. Plasma levels of TC, LDL-C, apoA-I and apoB were significantly higher while TG were significantly lower in middle-aged subjects, whereas both populations displayed similar HDL-C values. CEC values were significantly different between adults and octogenarians in the unadjusted analysis. Notably, elderly subjects showed a markedly higher serum CEC compared to adult subjects (average CEC ratio over serum standard was 1.25 ± 0.19 compared to 0.96 ± 0.10; p<0.0001; Figure 4). This difference remained significant even after adjusting for the propensity score (p<0.0001).

DISCUSSION
In the present study we demonstrated that in healthy individuals aged 80 years or more (80-102 years), serum CEC does not associate neither with atherosclerotic burden in the coronary arteries nor with features of plaque vulnerability. In addition, no relationship was detected between serum CEC and TL. In addition, as a secondary endpoint, we made a comparison with a cohort of middle-aged subjects, finding that octogenarians present higher CEC.
It is important to note that old subjects enrolled in our study, beyond the lack of CVD, bare a good nutritional status and have no evidence of neoplastic disease. This thorough selection allowed to avoid potential confounding factors, whose impact on cardiovascular risk factors is well documented (35).
In our previous paper involving a wider group of octogenarian individuals, we demonstrated that HDL-C independently associates with severity of subclinical coronary artery disease, whereas LDL-C does not (6). The current study adds new information, indicating that in very old subjects HDL function, measured by CEC, is not related to traditional cardiovascular risk factors, such as CAC or characteristics of vulnerable plaque. In addition, for the first time, we examined the association between serum CEC and TL, but we found no relationship.
Serum CEC is a metric of HDL functionality that has been associated to atherosclerosis and cardiovascular diseases in both cross-sectional and longitudinal manner, indicating that this parameter may be used as a valid cardiovascular risk biomarker, beyond plasma HDL-C concentrations (36,37). Notably, in most published CEC evaluations the apoB-depleted serum (HDL fraction) has been utilized, while in this study we used whole serum.
This could be considered a limitation in the assessment of HDL functionality. Indeed, in sera with cholesterol levels in normal range, which is the case of ours, previous observations indicate that cholesterol efflux from cAMP-treated macrophages to both whole and apoB-depleted serum normally leads to similar results (38,39). In our cohort we did not evidence differences of CEC values in subjects with coronary artery calcium score = 0 or > 0, although a not significant tendency of lower serum CEC in subjects with coronary artery calcium score > 0 was revealed. A further stratification by CAC tertiles revealed that subjects in the lowest tertile of CAC exhibit the same CEC as those in the highest tertile of CAC. Surprisingly, individuals in the intermediate tertile of CAC showed lower CEC than those in the lowest and the highest tertiles. We believe this pattern may represent a play of chance due to the limited number of subjects in the study. Previous works examined the association between CEC and coronary calcium: one recent study reported a reduction of HDL CEC in subjects with prevalent CAC (>0) (20), while other authors did not find any significant association between the two parameters (17). In both studies the analyzed population consisted in middle-aged subjects, thus the results may not reflect the situation in subjects over 80 years of age. Our data play in favor of an absence of a relationship between HDL functionality and CAC, which for the first time has been evaluated in individuals aged of 80 years or more. However, since the post-hoc power analysis (β=39%) was not sufficient to discard a potential mild difference between groups, this finding certainly deserves re-evaluation in future, larger studies. The lack of association between CAC and CEC could be explained by confounding factors differently affecting these parameters. Probably the main element to be considered is the divergence between the cumulative nature of the CAC and the dynamic nature of the CEC. While the former corresponds to the last 80 years or more of these individuals, CEC may reflect the function of HDL at a shorter period of time. Lifestyle and medication use during years may have had a stronger impact on atherosclerotic burden than HDL function, distancing a potential association between the two parameters. Statins, for example, are known to increase CAC (40), without affecting CEC (41,42). Although in our cohort, statin therapy was not related neither to CAC (median and range interquartile: 17 (0.0-763.5) and 119.5 (19.25-811.3) in users and non-users respectively; p = 0.312) nor to CEC (mean ± SD: were 1.45 ± 0.15 and 1.41 ± 0.17 in users and non-users respectively; p = 0.402), the present data is not sufficient to rule out this potential interaction.
The observation that the octogenarians present significantly higher CEC compared to younger individuals are in contrast with a previous report from Berrougui and colleagues. These authors compared CEC of two small cohorts, demonstrating that aging promotes phenotypic changes of HDL and impairment of their capacity to promote ABCA1-mediated cholesterol efflux (43), particularly the one driven by the smaller HDL 3 subpopulation. In addition, Berrougui showed that, differently from HDL 3 , the more mature particles HDL 2 from elderly subjects did not reveal an impaired function. We did not perform a structural characterization of HDL subpopulations in this present work, but based on Berrougui's findings, we may hypothesize that the HDL 2 contribution to cholesterol efflux may compensate for the HDL 3 impairment, possibly justifying the improved CEC observed in our subjects. In addition, in Berrougui's work the isolation of HDL was performed with ultracentrifuge that caused the elimination of nascent pre-beta HDL particles, as the authors themselves state, that are well known efficient acceptors of cholesterol released through ABCA1 and ABCG1 (44,45), both expressed in the cellular model that we have adopted (18). Conversely, in our samples it is conceivable that prebeta HDL significantly contribute to increased CEC by these mechanisms.
According to our results, we are tempting to speculate that improved CEC may represent a beneficial process, protecting elderly individuals from cardiovascular events, thus possibly explaining the lack of relationship with atherosclerosis in these subjects. However, a conclusive statement should be supported by the comparison with CEC values of octogenarian subjects with cardiovascular disease, as well with CEC values of younger descendants. We are well aware that a major limitation of the comparison of the two populations included in this study is the different ethnicity. Although mostly of these white elderlies are first and second generation European descendants who migrated to Brazil in the last century, we can not rule out that the observed difference is related, at least in part, to this. At this regard, up to now little is known about the impact of race on serum HDL CEC. A recent study specifically designed to compare HDL functionality in South Asians and Caucasians failed to demonstrate any difference, whereas Rohatgi and collaborators, by investigating the epidemiology of CEC in a large cohort from the Dallas Heart Study, observed that CEC was significantly lower in blacks than whites (17). Further, properly designed studies are warranted to clarify this point. In the present study we utilized CAC scoring as a risk-stratification tool because it is known to be useful in both middle-aged and elderly subjects. In fact, it has been shown that also in the elderly, elevated CAC is an independent predictor of cardiovascular events (46). However, also telomere shortening, considered a marker of biological aging, has been linked to age-related diseases such as atherosclerosis and identified as a predictor of its clinical outcomes (32). We did not find a correlation between TL and the markers of subclinical atherosclerosis flow-mediated dilation and intima-media thickness, in line with previous findings (32). In our population, we did not find an association between TL and coronary calcification as well. This is likely to be in contrast with the literature, where a relationship between these two parameters has emerged (47,48). However, such relationship has never been studied in the elderly, thus it is possible that, at very old age, similarly to what occurred for TC, the predictive power of TL is lost, further strengthening the concept that identifying risk factors in this specific population is difficult. For this reason, further studies conducted in patients are needed to deeper investigate on this aspect.
In our subjects, even the association between TL and outcomes of plaque remodeling is inconsistent. Whether this observation is related to the characteristics of enrolled subjects or reflects the absence of inter-relationship among these outcomes is still to be clarified. The former hypothesis could be supported by the observation that aging drives the loss of well established associations among cardiovascular outcomes (5). The latter could be investigated by evaluating CEC, TL and remodeling indexes in different cohorts of patients. To the best of our knowledge, these associations have not been evaluated yet neither in middle-aged subjects nor in elderly patients with chronic diseases.
In conclusion, similarly to the report by Mutharasan et al (49), our study failed to associate the atheroprotective parameter of serum CEC with indexes of atherosclerosis in very old subjects. As also suggested by these authors, it is possible to speculate that CEC plays a more protective role in the initial stages of atherosclerosis, thus explaining why the associations of CEC with subclinical atherosclerosis and cardiovascular events are more evident in younger populations. However, our original observation that people reaching old age in healthy conditions seems to have significantly higher CEC compared to a healthy, middle-aged population, offers a hint of great interest. Whether CEC may represent the cause or simply a marker of such a healthy longevity needs to be further investigated.

VARIABLE ELDERLY SUBJECTS (N=59)
Normally distributed data are presented as mean ± standard deviation and non-normally distributed data are presented in the form of median (interquartile range).