Association of Colony Forming Units with Coronary Artery and Abdominal Aortic Calcification

Abstract

Background

Certain bone marrow-derived cell populations, termed endothelial progenitor cells (EPCs), have been reported to possess angiogenic activity. Experimental data suggest that depletion of these angiogenic cell populations may promote atherogenesis, but limited data are available regarding their relation to subclinical atherosclerotic cardiovascular disease in humans.

Methods and Results

We studied 889 participants of the Framingham Heart Study who were free of clinically apparent cardiovascular disease (mean age, 65 years; 55% women). Participants underwent EPC phenotyping using an early outgrowth colony forming unit (CFU) assay and cell surface markers. Participants also underwent non-contrast multidetector computed tomography to assess the presence of subclinical atherosclerosis, as reflected by burden of coronary artery calcification (CAC) and abdominal aortic calcification (AAC). In this study sample, we examined the association of EPC-related phenotypes with both CAC and AAC. Across decreasing tertiles of CFU, there was a progressive increase in median CAC and AAC scores. In multivariable analyses adjusting for traditional cardiovascular risk factors, each standard deviation increase in CFU was associated with an approximately 16% decrease in CAC (P=0.02) and 17% decrease in AAC (P=0.03). In contrast, neither CD34+/KDR+ nor CD34+ variation were associated with significant differences in coronary or aortic calcification.

Conclusion

In this large, community-based sample of men and women, lower CFU number was associated with a higher burden of subclinical atherosclerosis in the coronary arteries and aorta. Decreased angiogenic potential could contribute to the development of atherosclerosis in humans.

BACKGROUND

Certain peripherally-circulating cell populations are believed to have endothelial reparative and angiogenic properties and, thus, have been termed endothelial progenitor cells (EPCs).¹ Experimental studies indicate that EPCs, broadly defined, are capable of promoting neovascularization in the setting of arterial ischemia^2,3 and re-endothelialization following mechanical arterial injury.⁴ Furthermore, EPC-related traits have been associated with cardiovascular risk factors in some clinical studies,^5–8 and data from selected samples suggest that decreased EPC number is related to an increased risk for adverse cardiovascular outcomes.^9,10 Thus, it has been hypothesized that chronic depletion of EPCs could contribute to the development of atherosclerosis.

Prior investigations of the association between circulating angiogenic cell phenotypes and subclinical vascular disease have been predominantly limited to small referral samples. In studies involving fewer than 100 subjects each, lower quantity of various EPC-related phenotypes has been associated with abnormal flow mediated dilation^6,11–14 and greater arterial stiffness.¹⁵ However, studies that have directly measured the presence and burden of subclinical atherosclerosis have not consistently demonstrated association with EPC traits.^{7,8,11,16–22} Data are particularly scant as well as conflicting with respect to measures of advanced subclinical atherosclerosis, such as coronary artery calcification (CAC).^8,19 If EPCs are critical for both maintaining endothelial integrity and facilitating arterial repair, chronically low EPC supply could predispose to atherosclerotic plaque formation. The presence of calcified plaque in the coronaries^23–25 and also the aorta^26,27 is considered a reliable marker of atherosclerosis and strongly predicts future cardiovascular events.²³ Thus, an association of EPCs with arterial calcification would support the hypothesis that EPC depletion contributes to the progression from subclinical endothelial dysfunction to cardiovascular disease in humans. We assessed EPC-related traits in a large, community-based sample and investigated whether or not low EPC quantity was associated with measures of coronary and aortic arterial calcification detected by multi-detector computed tomography (MDCT).

METHODS

Study Sample

In 1948, the Framingham Heart Study enrolled 5,209 residents of Framingham, MA in a longitudinal cohort study designed to identify risk factors for cardiovascular disease.²⁸ In 1971, a total of 5,124 offspring of the original cohort (and their spouses) were enrolled in the Framingham Offspring Study.²⁹ All participants in the Offspring study receive routine examinations approximately every 4 years, and a total of 3,021 participants attended their eighth examination cycle (2005 through 2008). Of this sample, 997 had phenotyping of EPC-related traits at the eighth examination and also had undergone assessment of coronary artery calcification (CAC) and abdominal aortic calcification (AAC) by MDCT between the seventh and eighth examinations (2002 through 2005). The MDCT scans occurred a mean of 2.6 (±0.9) years before the assessment of EPC phenotypes, which occurred during the eighth examination visit. The majority (n=889, 89%) of these participants did not have known cardiovascular disease (history of myocardial infarction, coronary insufficiency, stroke, and heart failure), and were eligible for the current analysis. All participants gave informed consent, and the institutional review board of the Boston University School of Medicine approved all study protocols.

Clinical and Risk Factor Assessment

All study participants underwent a standardized medical examination and laboratory assessment of cardiovascular risk factors. Systolic and diastolic blood pressures were the average of two physician-measured readings. Body mass index was calculated as weight divided by height squared (kg/m²). Blood was drawn for glucose, total and high-density lipoprotein (HDL) cholesterol, and triglycerides after overnight fast. Use of medications and cigarette smoking (current smoker within the prior year, past smoker, never smoker) were self-reported. Diabetes was defined as having a fasting glucose ≥126 mg/dL or taking medications to treat diabetes. Alcohol use was defined as ingestion of >14 or >7 alcoholic beverages per week in men or women, respectively. Physical activity was assessed using a physical activity index, calculated from the number of hours spent each day at various activity levels, weighted according to the estimated oxygen consumption required for each activity.³⁰ Education level was self-reported and assessed as a categorical variable (less than high school completion, high school diploma or equivalent but no college degree, completed college degree or higher). C-reactive protein (CRP) was assayed using the immunoturbidimetric latex-enhanced high sensitivity assay (Roche Diagnostics).

Assessment of Cell Phenotypes

Fasting blood specimens were collected from participants in the morning between 8 and 9 A.M. for assaying the following angiogenic cell phenotypes: colony forming unit (CFU), CD34+ cells, and CD34+/KDR+ cells. Following initial centrifugation of each blood specimen, the resulting buffy coat was further processed for cell phenotyping within 4 hours of specimen collection as previously described,^6,31 with modifications. Specifically, buffy coat samples were diluted to 10.5 mLs with PBS (Invitrogen) and layered over 5 mLs of Ficoll (Amersham Pharmacia Biotech). Each specimen was then centrifuged at 2200 rpm for 15 minutes at 10 degrees Celsius. Peripheral blood mononuclear cells were isolated from the buffy coat, using Ficoll density-gradient centrifugation, and then processed for CFU assay and flow cytometry.

Colony Forming Unit Assay

Collected peripheral blood mononuclear cells were washed with PBS, and then lysed with ACK lysis buffer (Fisher Scientific). Viable mononuclear cells, totaling 5 million per specimen, were then plated in each well of a 6 well fibronectin coated tissue culture plate (BD Biosciences) in M199/20% FBS and cultured at 37 degrees Celsius/5% CO2. Non-adherent cells were collected after 2 days, and 2 million viable cells in M199/20% FBS were re-plated in wells of a 24-well fibronectin coated tissue culture plate (BD Biosciences). After non-adherent cells were cultured for an additional 5 days, the number of newly formed colonies in each well was counted. Each distinct colony was identified using specific morphologic characteristics that have been previously described.⁶ A single, blinded technician performed colony counting and reported counts as the average number of colonies per well across up to 12 wells. In wells where the number of colonies was too numerous to count (mean of 4.7 wells from 63 individuals), the number of colonies per well was censored at 300. A single technician (RPM) performed initial plating of cells and colony counting for all specimens except for a small subset (KSC). Re-plating of cells was performed by one of two technicians (RPM, RJK). To minimize the effects of operator variation, colony counts were standardized by identity of the replater.

Flow Cytometry

Peripheral blood mononuclear cells were incubated for 15 minutes with FcR blocking agent (Miltenyi Biotec) on ice, and then for 25 minutes on ice with anti-KDR PE(R&D Systems) and anti-CD34 FITC(BD) anti-human antibodies. Samples were washed and then fixed in 2% paraformaldehyde. Expression of the surface markers was evaluated by fluorescence-activated cell sorter (FACS) analysis; the number of positive cells was quantified using a Becton-Dickinson FacsCalibur flow cytometer, with fluorochrome-matched IgG isotype controls. Red blood cells, platelets, and cell debris were excluded using forward and side scatter electronic characteristics. The frequency of CD34+ cells was then identified within the nucleated cell gate using population gating. Finally, KDR+ events within the CD34 population were analyzed via population gating. CD34+ and CD34+/KDR+ cells were quantified using FlowJo analysis software (Treestar).³¹ Quantities of each cell type were reported as percentage of the total number of gated nucleated cells. All flow analysis plots were reviewed by a blinded investigator (KSC) to ensure consistency.

Assessment of Arterial Calcification

Imaging of the chest and abdomen using an 8-slice MDCT (Lightspeed Ultra, GE, Milwaukee, WI, USA) scanner was performed for all participants, as previously described.³² Two chest scans and 1 abdominal scan were performed for each participant using a sequential scan protocol with a slice collimation of 8 mm × 2.5 mm (120 KVp, 320/400 mA for a cutpoint of 220 lbs body weight, respectively) during a single end-inspiratory breath hold (typical duration 18 seconds). Image acquisition (330 ms) was prospectively initiated at 50% of the cardiac cycle. For abdominal scanning, 30 contiguous 5 mm thick slices of the abdomen were acquired, covering 150 mm above the level of S1. A calibration phantom (Image Analysis, Lexington, KY, USA) containing rods of water and 75 and 150 mg/cm³ calcium hydroxyapatite, was placed underneath each participant.

A trained technician performed calcium measurements for each study using an offline workstation (Acquarius, Terarecon, San Matteo, CA, USA). The method for scoring CAC has been described previously, along with excellent intra- and inter-reader reproducibility for CAC measurements.³² A calcified lesion in either the coronary arteries or in the aorta was defined as an area of ≥3 connected pixels with CT attenuation >130 Hounsfield Units using 3D connectivity criteria. A score for AAC (from the abdominal scan) and CAC (from each of the 2 chest scans) was calculated by multiplying the area of a calcified lesion by a CT attenuation score weighted based on the maximal CT attenuation (Hounsfield Units) within a lesion. Because the original Agatston Score was originally developed for electron beam CT,³³ we applied a modified Agatston Score algorithm to our MDCT scan protocol to score for CAC as well as for AAC, as has been done in numerous prior studies.^32,34

Statistical Analyses

Due to highly skewed distributions, natural logarithmic transformed values were applied for triglycerides, CD34+/KDR+, and CD34+. Square-root transformed values were used for CFU after standardization by technician replater. We used natural logarithmic transformed values for CAC and AAC, where log(CAC+1) and log(AAC+1) were used in analyses, respectively, to account for zero values.

Associations of each measure of arterial calcium (CAC and AAC) with each cell phenotype (CFU, CD34+, and CD34+/KDR+) were characterized using linear regression models adjusting for age and sex as independent covariates. We also analyzed the association of CFU quantity with increasing number of traditional cardiovascular risk factors, defined as older age (>45 years for men, >55 years for women), male sex, hypertension, hypercholesterolemia, diabetes, current smoker, and obesity. Multivariable regression models adjusted for clinical covariates known to correlate with subclinical vascular disease: age, sex, body mass index (BMI), total/HDL cholesterol ratio, log triglycerides, smoking, systolic blood pressure, treated hypertension, treatment with lipid-lowering medication, diabetes, alcohol use, education level, and physical activity. We used generalized estimating equations,³⁵ with exchangeable correlation structure, to accommodate non-normalized correlated sibling data in families of different sizes with robust (sandwich) covariance estimators.

In secondary analyses, we repeated multivariable analyses including further adjustment for CRP. We also tested for effect modification by age or sex using multiplicative interaction terms in the multivariable models. All analyses were performed using SAS statistical software (GENMOD procedure), version 9.1.3. A two-tailed P value of <0.05 was considered significant.

RESULTS

Clinical characteristics of the study sample are shown in Table 1. The mean age was 65, and 55% were women. The median (and interquartile range) of CFU number, CD34+ %, and CD34+/KDR+ % were 41 (20–62), 7.7 (4.9–10.5), and 0.3 (0.15–0.45) in the total sample, respectively. CFU quantity was not correlated with log CD34+ (age- and sex-adjusted Pearson correlation coefficient r=0.06, P=0.12) or log CD34+/KDR+ (r=0.05, P=0.20). Log CD34+ and log CD34+/KDR+ were moderately correlated (r=0.26, P<0.001). Estimated sibling correlations were 0.124 for log (CAC+1) and 0.236 for log (AAC+1).

Table 1.

Sample characteristics

	Total Sample	1st CFU tertile	2nd CFU tertile	3rd CFU tertile
	N=889	N = 258	N = 257	N = 258
Age, year	65±9	66±9	65±8	64±9
Women, %	55	61	56	47
Body mass index, kg/m2	28.4±5.3	28.0±5.1	29.2±5.8	28.2±5.0
SBP, mmHg	128±16	128±17.	129±17.	127±15.
DBP, mmHg	74±10	74±10	75±10	75±9
Hypertension, %	58	58	65	53
Hypertension medications, %	48	47	54	45
Total cholesterol, mg/dL	188±36	191±35	189±38	183±32
HDL cholesterol, mg/dL	57±17	58±17	58±18	56±16
Total/HDL cholesterol	3.5±1.0	3.5±1.0	3.5±1.0	3.5±1.0
Triglycerides, mg/dL*	101 (67–135)	100 (94–107)	102 (73–146)	97 (71–133)
Cholesterol meds, %	46	42	47	49
Fasting glucose, mg/dL	105±21	104±20	105±18	106±27
Diabetes, %	10	10	10	9
Smoking status, %
Current	5	5	5	5
Past	60	59	62	60
Never	35	36	33	35
Alcohol use,† %	16	16	18	14
Physical activity index	35.5±5.6	35.7±5.7	35.5±5.7	35.6±5.5
Education level, %
Less than high school	2	2	3	3
High school diploma	56	54	58	55
College degree	42	44	39	42

SBP, systolic blood pressure; DBP, diastolic blood pressure.

^*Values presented as median (interquartile range) for non-normally distributed variables. All other values are presented as means ± standard deviations or percentages.

^†Defined as >14 drinks per week (men) or >7 drinks per week (women).

Individuals with lower CFU had higher CAC and AAC scores. Median CAC scores were 15,0, 22.2, and 34.7 Agatston units, across decreasing CFU tertiles. Corresponding AAC scores were 274.1, 472.0, and 498.1 Agatston units. In this sample, lower CFU number was not associated with increasing number of traditional cardiovascular risk factors.

The results of the multivariable regression analyses are shown in Table 2. There were significant inverse relations of CFU with both CAC and AAC after adjustment for age, sex, and conventional cardiovascular risk factors and accounting for between-sibling correlations. An SD increment in the CFU variable was associated with a −0.177 decrease in log CAC (95% confidence intervals [CI] −0.320, −0.034), which corresponds to an approximately 16% lower CAC score. Each SD increment in CFU was also associated with a −0.185 decrease in log AAC (95% CI −0.356, −0.013), corresponding to an approximately 17% lower AAC score. In contrast, neither CD34+ nor CD34+/KDR+ was associated with CAC or AAC in multivariable analyses (Table 2).

Table 2.

Results of linear regression models assessing the relation of CT-measures of calcification to CFU, CD34+, and CD34+/KDR+ cell phenotypes

Independent Variables	Age- and Sex-Adjusted		Multivariable-Adjusted†
	Coefficient (95% CI)	P-value	Coefficient (95% CI)	P-value
Coronary arterial calcification (as dependent variable)
CFU	−0.155 (−0.303, −0.007)	0.04	−0.177 (−0.320, −0.034)	0.02
CD34+	0.014 (−0.294, 0.322)	0.93	−0.028 (−0.191, 0.135)	0.74
CD34+/KDR+	0.020 (−0.177, 0.217)	0.84	0.024 (−0.126, 0.174)	0.75
Abdominal aortic calcification (as dependent variable)
CFU	−0.160 (−0.328, 0.009)	0.06	−0.185 (−0.356, −0.013)	0.03
CD34+	0.336 (−0.031, 0.702)	0.07	0.132 (−0.047, 0.312)	0.15
CD34+/KDR+	−0.007 (−0.227, 0.213)	0.95	0.008 (−0.152, 0.168)	0.92

CI, confidence intervals; CFU, colony forming unit.

^*Regression coefficients represent change in the dependent variable per standard deviation change of square-root CFU, log CD34+, and log CD34+/KDR+, respectively, in analyses that accounted for correlated sibling data using generalized estimating equations with exchangeable correlation structure.

^†Covariates adjusted for in the multivariable model include: age, sex, body mass index, total/HDL cholesterol, log triglycerides, smoking status, systolic blood pressure, taking medication for hypertension, taking medication for hypercholesterolemia, diabetes, alcohol use, education level, and physical activity.

In secondary analyses, the associations of CFU with CAC and AAC were unchanged in multivariable analyses that additionally adjusted for CRP (P=0.01 and P=0.04 for associations of CFU with CAC and AAC, respectively). We performed analyses testing for effect modification by age and sex. For AAC, there was a statistically significant interaction of age and CFU (P=0.004), with individuals <65 years having a stronger association between CFU and AAC (β −0.365, 95% CI −0.634, −0.095) than those ≥65 years (β 0.097, 95% CI −0.096, 0.290). In contrast, the interaction between age and CFU was non-significant for CAC (P=0.18; younger individuals, β −0.240 [95% CI −0.429, −0.050]; older individuals, β −0.054 [95% −0.277, 0.170]). There was no evidence of a sex interaction for the association of CFU with either measure of arterial calcification.

DISCUSSION

In a large, community-based sample of predominantly healthy men and women, reduced CFU quantity was associated with greater subclinical coronary and aortic calcification. This association was present even after adjusting for cardiovascular risk factors. These findings are consistent with the hypothesis that reduced angiogenic potential could contribute to the development of atherosclerosis.

On the basis of experimental data, it has been proposed that chronic depletion of circulating angiogenic cells, typically referred to as EPCs, leads to an impaired capacity for endothelial repair. Clinical studies in selected samples have related EPC traits, using either culture-based assays or flow cytometry, with endothelial function and arterial compliance.^{6,11,12,14,15,36} However, data regarding the relation of EPCs with measures of subclinical atherosclerosis have been mixed. In one of the largest of prior studies, which included 137 middle-aged men and women, Fadini and colleagues reported an association of carotid intima-media thickness with CD34+/KDR+ but not CD34+ cell populations.¹⁷ Other studies, however, have found no association between carotid intima-media thickness and CD34+/KDR+.^11,16 Coronary calcification generally reflects the presence of more advanced subclinical atherosclerosis, and is a strong predictor of future cardiovascular events.^23–27 In a sample of 90 healthy men, Bielak and colleagues reported on the inverse relation of CAC with CD34+ cells but did not examine its relation with other EPC phenotypes.⁸ In contrast, Hughes and colleagues found no association between CD34+ cells and CAC among 117 European and South Asian men.¹⁹

To our knowledge, the present study is the largest clinical investigation involving EPC-related traits. Furthermore, our investigation focused on ambulatory individuals without acute illnesses that could lead to mobilization of pro-angiogenic cells.^37,38 We observed a significant association of lower CFU with higher CAC and AAC scores. In contrast, we did not observe an association between CD34+ or CD34+/KDR+ and arterial calcification. These apparently discordant findings are consistent with experimental studies suggesting that CFU is a distinct cell-based phenotype from CD34+ and CD34+/KDR+ with respect to both character and function.^1,39

CFUs are comprised of hematopoietic cells that produce large quantities of angiogenic cytokines and enhance assays of vessel formation both in vitro and in vivo.⁴⁰ The formation of CFUs appears to be dependent on lineage restricted cell populations, predominantly monocytes and T-cells.⁴¹ The dependence of CFU formation on monocyte populations may have particular relevance for their observed association with CAC and AAC. In experimental models, certain monocyte subsets have been observed to promote anti-inflammatory rather than pro-inflammatory activity, in addition to angiogenesis and granulation tissue formation.⁴² Monocyte-derived cells have also been associated with remodeling and regression of arterial calcification.⁴³ Thus, based on their affiliation with the monocytic lineage, it is plausible that the CFU phenotype reflects cellular activities that directly affect the progression or regression of atherosclerotic lesions and, particularly, calcific lesions in the setting of advanced subclinical disease. This effect may be more evident in younger individuals, given the higher prevalence and severity of other factors unrelated to angiogenesis in older individuals that promote atherosclerosis and vascular calcification.⁴³

In contrast to CFUs, the CD34+ and CD34+/KDR+ phenotypes are largely comprised of hematopoietic stem/progenitor cells as well as a smaller population of cells with late-outgrowth endothelial colony formation. Accordingly, prior studies have also shown poor correlation between CD34+ related phenotypes and CFUs (which can form in the absence of CD34+ cells).¹ CD34+ and KDR+ phenotypes may lack an observed association with arterial calcification for several reasons. First, CD34+ and CD34+/KDR+ cells may not be physiologically active in processes related to vascular calcification. Since arterial calcification denotes the presence of advanced atherosclerotic plaque, it is possible that CD34+ and CD34+/KDR+ quantity reflects a capacity for endothelial repair that is associated with earlier rather than later manifestations of atherosclerosis, or alterations in vascular integrity that do not result in calcification. Second, CD34 and KDR cell surface markers, even when used together, are not highly specific for progenitor cells with endothelial repair and regenerative capacity.¹ CD34 is present not only on hematopoietic stem/progenitor cells, but also on mature endothelial cells and embryonic cells.^44,45 Although KDR, a type of VEGF-receptor 2, may identify a population of circulating hematopoietic cells with more specific endothelial characteristics,^10,45 some evidence suggests that even CD34+/KDR+ cells represent primitive hematopoietic rather than more differentiated endothelial progenitors.⁴⁶ Thus, the ability to identify a true association between biologically relevant progenitor cells and vascular calcification is reduced. Finally, the number of circulating cells positive for both CD34 and KDR is very low in ambulatory individuals,⁴⁷ which may limit statistical power for detecting significant effects. This could also account for differences between our results and those observed in higher risk populations.^5,9

Several limitations of this study merit consideration. Our analyses were cross-sectional, and thus we cannot determine whether low CFU quantity preceded or followed the development of coronary atherosclerosis. It is also possible that associations of CFU with CAC and AAC are related to unmeasured confounding factors that may influence both colony number and the development of atherosclerosis. In particular, extracellular factors could impact cellular characteristics that contribute to the regenerative capacity of EPCs (e.g. paracrine function, migration, resistance to stress, and senescence) and, in turn, result in both lower CFU quantity and increased cardiovascular risk. Because we examined several measures of EPC traits and arterial calcification, we cannot exclude the possibility of false positive findings due to multiple comparisons. Nonetheless, our findings are consistent with studies in referral samples relating CFUs to overt cardiovascular events.⁹ Analysis of endothelial function could provide additional information regarding factors mediating the observed relation of CFU with arterial calcification; however, concurrent assessments of endothelial function were not available for analysis in our study sample.

Despite their recognized angiogenic potential,⁴⁸ CFUs likely comprise a heterogeneous population of cells with potentially distinct physiologic properties.⁴⁹ Likewise, cells identified as CD34+ and CD34+/KDR+ are likely to include multiple cell subtypes, some of which may not exhibit angiogenic activity. Further stratifying CFUs, CD34+, and/or CD34+/KDR+ cells based on additional surface antigens, markers of senescence, and/or migratory capacity could provide additional insights.⁵⁰ Additionally, our sample was predominantly white and of European ancestry, limiting the generalizability of our findings to other racial/ethnic populations.

In summary, lower CFU number was associated with greater coronary and aortic calcification, consistent with a higher burden of subclinical atherosclerosis. In the context of prior studies relating CFU quantity with cardiovascular risk factors and adverse cardiovascular events, these findings suggest that CFUs may represent a distinct marker of circulating angiogenic potential that contributes to vascular health and protects against the pathophysiologic changes that promote atherosclerosis. Further research is required to investigate the potential mechanisms underlying the association of CFUs with arterial calcification and determine if, indeed, interventions that augment CFU quantity could be effective at preventing or treating clinical cardiovascular disease.