Abstract
Objective. Osteoprotegerin (OPG) may contribute to the link between systemic inflammation and increased cardiovascular risk. We investigated the relationship of OPG concentrations with endothelial activation and carotid atherosclerosis in rheumatoid arthritis (RA).
Methods. OPG concentrations and those of endothelial activation molecules were measured by using ELISA in 34 patients who were treated with infliximab (IFX), both immediately before and after an IFX infusion. Carotid intima-media thickness (CIMT) and plaque were determined by ultrasound in 27 of the study participants.
Results. Median (interquartile range) OPG concentrations decreased from 4.8 pmol/l (2.8–6.5) to 4.4 pmol/l (2.9–6.1; p = 0.04) upon IFX infusion. Baseline OPG concentrations were inversely associated with those of total and low-density lipoprotein (LDL) cholesterol (partial R = −0.50, p = 0.004, and R = −0.48, p = 0.007, respectively). Prior to IFX administration, OPG concentrations were associated with those of intercellular adhesion molecule (ICAM)-1 (partial R = 0.34, p = 0.05), CIMT (partial R = 0.51 to 0.52, p < 0.009), and plaque (OR = 1.52, 95% CI 1.01–2.29 to OR = 1.61, 95% CI 1.03–2.51; p < 0.04), independent of conventional risk factors and C-reactive protein concentrations or disease activity. Except for the OPG concentrations-plaque association (p = 0.09), these relationships remained significant subsequent to IFX administration (p < 0.05). Reductions in OPG levels related to those in vascular cell adhesion molecule (VCAM)-1 concentrations (partial R = 0.35, p = 0.04) and had borderline significance (p = 0.09) with those in ICAM-1 (partial R = 0.29) concentrations.
Conclusion. OPG concentrations are independently associated with endothelial activation and carotid atherosclerosis in RA. Reductions in OPG concentrations upon IFX administration are associated with decreased endothelial activation. OPG may be involved in increased cardiovascular disease risk and may improve its stratification in patients with RA.
Osteoprotegerin (OPG) is a member of the tumor necrosis factor (TNF) receptor superfamily, plays a central role in bone remodeling, and acts as a decoy receptor for the receptor activator of nuclear factor-κB ligand (RANKL)1,2. OPG thereby reduces binding of RANKL to its receptor, RANK. The OPG/RANKL/RANK axis is strongly involved in osteoporosis and increasingly implicated in atherosclerosis1,2. In addition, OPG was documented to act as a soluble neutralizing receptor for another member of the TNF superfamily, TNF-related apoptosis-inducing ligand (TRAIL), which is a molecule that exhibits antiinflammatory and antiatherosclerotic properties3,4,5. OPG is produced by most human tissues and cells that include not only osteoblasts, but also endothelial and vascular smooth muscle cells. Proinflammatory cytokines enhance OPG production and OPG may also participate in the pathophysiology of rheumatoid arthritis (RA)6,7. Moreover, C-reactive protein (CRP) can downregulate the expression of TRAIL4.
OPG knockout mice exhibit severe osteoporosis, as well as arterial media calcification, indicating a potential protective effect of OPG against calcification2,8. However, this phenotype of arterial calcification occurs in the absence of fat deposition or atherosclerotic-like plaques and does not involve intimal calcification as observed in atherosclerosis. Further, cellular studies revealed that OPG can upregulate endothelial adhesion molecule production9 and indeed increase leukocyte adhesion to endothelial cells10. OPG is present in atherosclerotic plaques11. Thus, OPG may have beneficial, detrimental, or dual effects on atherosclerosis development and progression2.
Human studies have consistently revealed positive relationships between circulating OPG concentrations and an increased prevalence and severity of coronary artery disease12,13, incident coronary artery disease14,15, cerebrovascular disease16, and peripheral vascular disease17. In view of the findings in OPG knockout mice, it was postulated that the reported associations between OPG concentrations and cardiovascular disease (CVD) in humans may represent paradoxical and compensatory increases in OPG production in an attempt to attenuate atherogenesis in the presence of established CVD15.
Patients with RA experience a markedly enhanced traditional and nontraditional risk factor-mediated CVD burden18,19,20,21,22,23. In RA, determination of increased OPG concentrations could be a promising candidate in enhancing CVD risk stratification as it may contribute to the link between high-grade inflammation and enhanced atherogenesis. A positive relationship between OPG concentrations and coronary artery calcification was recently documented in patients with RA24. However, OPG-related increased coronary artery calcification may not necessarily represent atherosclerosis2,8. Also, coronary artery calcification scores are far less sensitive than ultrasound (US)-determined carotid artery atherosclerosis in identifying patients at high risk of CVD25,26. Recently, an age, sex, and CRP concentration-independent association between OPG levels and carotid artery plaque was documented in RA27. Whether OPG concentrations associate with endothelial activation in RA is, to our knowledge, unknown at present. Taken together, the role of OPG production in enhanced atherogenesis and CVD risk stratification in RA requires further elucidation. Therefore, in our study, we examined whether OPG concentrations are associated with those of endothelial activation molecules, as well as carotid intima-media thickness (CIMT) and plaque, independent of conventional cardiovascular risk factors, systemic inflammation, and disease activity in a cohort of patients with severe RA. We also determined whether upon the administration of the TNF-α blocker infliximab (IFX), potential reductions in OPG concentrations6 were related to decreased endothelial activation.
MATERIALS AND METHODS
Patients
Thirty-four patients who met the 1987 American College of Rheumatology28 and 2010 American College of Rheumatology/European League Against Rheumatism criteria for RA29 participated in our study. Each patient used IFX, 3 to 5 mg/kg, at 6 to 8 weekly intervals for at least 1 year, and oral methotrexate 15 to 25 mg weekly, with or without chloroquine (250 mg/day); prednisone 2.5 to 7.5 mg daily; and a nonsteroidal antiinflammatory agent. Also, 59.9% and 20.6% of participants had been treated with leflunomide and sulfasalazine, respectively, before the initiation of IFX therapy. The local institutional committee approved anti-TNF-α therapy and each patient gave informed consent to participate in the study.
Methods
We recorded demographic features, cardiovascular risk factors, cardiovascular agent use, disease characteristics, and markers of systemic inflammation. All measurements were taken in a fasting state. The Disease Activity Score in 28 joints (DAS28)30 was calculated. Blood samples were taken at 8 a.m. to measure the erythrocyte sedimentation rate and concentrations of CRP (latex immunoturbidimetry), lipids (enzymatic colorimetry, plasma glucose, and insulin; DPC, Dipesa). Insulin resistance was estimated by the homeostasis model of insulin resistance.
As reported31, the concentrations of the soluble endothelial activation molecules, including E-selectin, P-selectin, vascular cell adhesion molecule (VCAM)-1, intercellular adhesion molecule (ICAM)-1, and ICAM-3, were determined by ELISA (Bender MedSystems) before and 120 min after an infliximab infusion; their intraassay coefficients of variation and lower detection limits were 5.4% and 0.5 ng/ml, 2.4% and 1.06 ng/ml, 3.1% and 0.9 ng/ml, 4.1% and 3.3 ng/ml, and 2.5% and 0.58 ng/ml, respectively.
Carotid US was performed in 27 (79.4%) of participants. Our methodology used to assess carotid atherosclerosis was also reported previously32. The right-sided CIMT was evaluated using QLAB’s CIMT-quantification software measurement plug-in (Philips Healthcare, DA Best). This instrumentation allows for highly consistent and reliable CIMT measurements, reducing the effort required to make assessments successfully and minimize the time needed to complete the respective investigations. For the purpose of our study, atherosclerotic plaque was defined as a distinct protrusion of > 1.5 mm into the vessel lumen33. We used a high-resolution B-mode US (iE33 Philips ultrasound system, Philips Ultrasound) with a 10-MHz linear transducer. A single cardiologist (CG-J) performed all measurements. The reproducibility of CIMT measurements was evaluated in 17 patients and 17 controls within 1 week of the first US examination. The correlation coefficient for CIMT was 0.98.
Using an ELISA (Biomedica), we determined OPG concentrations before and 120 min after an IFX infusion. The intraassay coefficient of variation and lower detection limit were 10% and 0.14 pmol/l, respectively. As reported, we also measured serum concentrations of 3 adipokines including adiponectin34, resistin35, leptin36, and ghrelin37 before and after the IFX infusion.
Data analysis
Results are expressed as median [interquartile range (IQR)] and proportions for continuous and categorical variables, respectively. Changes in OPG concentrations upon IFX infusion were assessed by the Wilcoxon Signed-Rank test. OPG concentrations in women and men were compared by the Mann-Whitney U test. Associations of OPG concentrations with cardiovascular risk factors, RA characteristics, markers of systemic inflammation, adipokine, ghrelin, and endothelial activation molecule concentrations, CIMT, and plaque were evaluated using potential confounder or/and mediating characteristics adjusted mixed linear or logistic regression models as appropriate.
Statistical computations were done using the GB Stat program (Dynamic Microsystems Inc.).
RESULTS
Recorded characteristics and changes in OPG concentrations upon IFX administration in 34 patients with RA
Table 1 shows the recorded characteristics in the 34 participants. Their median age was 57 years and 73.5% were women. Despite the use of conventional disease-modifying antirheumatic drug and TNF-α blockade in each patient, the overall disease activity was still in the moderate range with a median DAS28 of 4.3 (IQR: 3.3–5.1), and only 1 patient experienced disease remission (DAS28 < 2.4). Hence, patients included in our study had severe RA. The levels of each of the evaluated endothelial activation molecules decreased upon IFX administration; as previously recorded by us, these reductions were significant for ICAM-3 and P-selectin31. Importantly, these decreases were paralleled by a significant (p = 0.04) reduction in the median OPG concentrations from 4.8 (IQR: 2.8–6.5) to 4.4 (IQR: 2.9–6.1) pmol/l.
Age-adjusted and sex-adjusted associations of OPG concentrations
Concentrations were measured at time 0 and 120 minutes with cardiovascular risk factors, disease characteristics, systemic inflammation, and adipokine and ghrelin concentrations in patients with RA. Baseline and post-infusion IFX OPG concentrations were numerically higher in women than men [median = 5.1 (IQR: 2.9–6.9) vs 3.8 (IQR: 2.8–5.0) and 5.0 (IQR: 2.9–6.1) vs 3.7 (IQR: 2.8–4.4) pmol/l, respectively], but these differences were not significant (p = 0.4 and 0.2, respectively). In the sex-adjusted analysis, age at disease onset and at the time of the study were associated with baseline and post-IFX infusion OPG concentrations [partial R = 0.327 (p = 0.05) and 0.390 (p = 0.02), and 0.344 (p = 0.04) and 0.419 (p = 0.02), respectively].
Table 2 gives the age at the time of the study (replaced by age at disease onset for model that included disease duration) and sex-adjusted associations of baseline OPG concentrations with cardiovascular risk factors, RA characteristics, systemic inflammation markers, and adipokine and ghrelin concentrations in the 34 participants. OPG concentrations were inversely associated with those of total and LDL cholesterol. These relationships were materially unaltered upon further adjustment for CRP concentrations or DAS28 (data not shown). Further, OPG concentrations related to those of CRP with borderline significance (p = 0.09). OPG concentrations were not related to those of adipokines and ghrelin. Further adjustment for cardiovascular drug use in the models in Table 2 did not materially alter the findings (data not shown).
Independent relationships of baseline and post-IFX infusion OPG concentrations with endothelial activation and carotid atherosclerosis
Tables 3A and 3B show the independent relationships of OPG concentrations with endothelial activation and carotid atherosclerosis at baseline and after IFX administration, respectively. Independent of conventional risk factors as represented by the Framingham score and systemic inflammation as reflected by CRP concentrations, baseline OPG concentrations related to those of ICAM-1 and both CIMT and plaque; when the Framingham score and disease activity as estimated by the DAS28 were adjusted for, baseline OPG concentrations were associated with those of both VCAM-1 and ICAM-1 concentrations, and CIMT, as well as plaque. Post-IFX OPG concentrations were related to those of ICAM-1, CIMT, and plaque independent of conventional risk factors and systemic inflammation. These relationships were materially unaltered upon adjustment for conventional risk factors and disease activity, although the post-infusion OPG-carotid plaque association was only borderline significant (p = 0.09). In our previous analysis, among the conventional risk factors, only total and LDL cholesterol concentrations were associated with those of OPG, and the DAS 28 was also not related to OPG concentrations (Table 2). Upon reassessing the OPG-endothelial relations in Table 3A with adjustment for only total and LDL cholesterol, the results were materially unaltered [partial R = 0.35 (95% CI 0.00 to 0.70), p = 0.04 and partial R = 0.33 (−5% CI −0.02 to 0.68), p = 0.06, for the OPG-VCAM-1 and OPG-ICAM-1 relationships, respectively].
In view of our results as shown in Tables 3A and 3B, we determined whether changes in OPG concentrations were associated with changes in VCAM-1 and ICAM-1. Reductions in OPG levels related to those in VCAM-1 concentrations [partial R = 0.35 (95% CI 0.01 to 0.68), p = 0.04] and with borderline significance (p = 0.09) to those in ICAM-1 [partial R = 0.29 (95% CI −0.16 to 0.63)] concentrations.
Post-IFX infusion OPG concentrations were not related to those of post-infusion adipokines and ghrelin, and changes in OPG levels were also unrelated to those in adipokines and ghrelin34,35,36,37 (data not shown).
DISCUSSION
To our knowledge, our study shows for the first time that, in patients with severe RA, OPG concentrations relate to enhanced endothelial activation independent of well-established CVD risk factors, and that upon IFX infusion, reductions in OPG concentrations are associated with decreased endothelial activation. Patients with severe RA are particularly prone to increased cardiovascular risk38. Our results also confirm previously reported findings of an association between OPG concentrations and carotid atherosclerosis27 and that TNF-α blockade can reduce OPG concentrations in RA6. These findings indicate that increased OPG production is associated with markers of atherogenesis in RA. Finally, the favorable relationships between OPG concentrations and lipid measures, as produced by our data analysis, have been documented in subjects without RA39, but not in those with RA, and may therefore constitute a severe disease-specific feature among patients with RA.
Our investigation was partly cross-sectionally designed, which precludes drawing inferences on the direction of causality. Therefore, our finding of an independent relationship between OPG concentrations and those of endothelial activation molecules, as well as carotid atherosclerosis, may represent a proatherogenic effect or a compensatory increase in OPG production in the presence of heightened CVD risk2,15. Nevertheless, we also found that changes in OPG concentrations upon IFX administration were significantly associated with alterations in VCAM-1 concentrations. Taken together, our findings are congruent with those reported in cellular studies that assessed the effects of OPG on endothelial activation9,10. Notably, we also found a borderline significant relationship between CRP concentrations and those of OPG. Because CRP can further downregulate TRAIL expression and OPG acts as a decoy receptor for TRAIL3,4,5, the OPG-endothelial activation and OPG-atherosclerosis associations as found in the present investigation may also be partly attributable to an inhibition of TRAIL. Identification of the exact nature of the OPG-cardiovascular risk relationship requires further investigation to determine the potential role of OPG targeting in reducing CVD risk in RA.
Paradoxical associations between adipokine concentrations and CVD risk were reported in subjects without and with RA and may represent a compensatory mechanism in the presence of enhanced CVD risk and in an attempt to reduce atherogenesis40,41,42. Whether this phenomenon underlies the inverse association between OPG concentrations and those of total and LDL cholesterol in our investigation deserves further study. Indeed, OPG concentrations were significantly related to increased endothelial activation and atherosclerosis, even after adjusting for lipid measures, thereby arguing against an overall protective effect of OPG on CVD risk.
Patients with RA experience cytokine induced enhanced endothelial activation molecule production that further associates with prevalent and incident atherosclerosis43,44,45. Conceptually, enhanced endothelial activation could reflect RA activity46. However, disease activity and systemic inflammation were not associated with enhanced endothelial activation in the present investigation (data not shown) and the relationship between OPG concentrations and those of endothelial activation molecules were independent of both CRP concentrations and the DAS28.
Current recommendations on CVD risk stratification in RA38 may result in a substantial proportion of patients at high or very high risk for cardiovascular events remaining unidentified and hence receiving inadequate CVD management26. It was recently shown that carotid ultrasound may be particularly useful in improving CVD risk stratification in RA47. Another promising tool in this context is the use of biomarkers with enhanced atherogenesis, especially prior to plaque development, in RA26. Examples of this possibility include using resistin35,48 and angiopoietin-249 concentrations, as well as genetic markers50,51. We found that OPG concentrations related not only to carotid atherosclerosis27, but also endothelial activation independent of systemic inflammation and RA activity, as well as conventional CVD risk factors. Such an independent relationship was similarly shown between OPG concentrations and coronary artery calcification scores in RA24. Thus, our results together with previously reported findings support a potential role of OPG concentrations in CVD risk stratification in RA. Such findings further suggest that, as applies to resistin48, OPG may contribute to the link between systemic inflammation and increased CVD risk, and may assist in CVD risk stratification beyond information obtained through evaluation of routine systemic inflammatory markers.
In a previous study6, initiation of treatment with IFX in 21 patients with RA resulted in a gradual decrease in OPG concentrations that was significant after 6 weeks (but not within 2 weeks) after the first infusion and normalized after 14 to 22 weeks. Therefore, it is noteworthy that in our cohort of patients with severe RA that had already received IFX prior to the study, a significant reduction was observed within 2 h after a repeat infusion. We previously reported decreases in endothelial activation31 and resistin concentrations35 and enhancement in insulin sensitivity52 in our cohort. Each of these processes43,44,45,48,53,54 reportedly contributes to enhanced atherogenesis in RA. Taken together, IFX administration may rapidly result in favorable effects on cardiovascular risk in patients on periodic treatment with this drug.
Systemic inflammation associates with high resistin35,48 and low adiponectin concentrations34 and IFX therapy can also reduce ghrelin concentrations36 in RA. In our investigation, OPG concentrations were consistently unrelated to those of adipokines and ghrelin. If OPG on the one hand and adipokines and ghrelin on the other indeed contribute to altered cardiovascular risk in RA, they may effectuate this through different pathways.
Our present investigation has other limitations. Only 34 patients who had severe RA were included. Nevertheless, OPG concentrations were associated with both endothelial activation and carotid atherosclerosis, and this was consistent before and after an IFX infusion. Serum OPG levels do not necessarily reflect tissue concentrations. Future RA studies may also benefit from including patients in disease remission as inflammation in active RA can be expected to associate with increased OPG concentrations. Whether OPG levels predict progression of CIMT and plaque also merits further study.
OPG is likely involved in enhanced cardiovascular risk in RA. The direction of causality within the OPG concentrations-endothelial activation and OPG concentrations-atherosclerosis relations needs additional elucidation in future mechanistic and longitudinal studies prior to targeting OPG in CVD risk management in RA. Meanwhile, determination of OPG concentrations may be useful in improving CVD risk stratification in RA.
Footnotes
-
Drs. Dessein and Gonzalez-Gay share senior authorship.
-
Supported by grants from “Fondo de Investigaciones Sanitarias” PI06/0024, PS09/00748, and PI12/00060 (Spain). This work was also partially supported by RETICS Program, RD08/0075 and RD12/0009/0013 (RIER) from “Instituto de Salud Carlos III” (ISCIII; Spain), and a Medical Research Council Grant (MRC2008_DES) from the National Research Foundation (South Africa; PH Dessein).
- Accepted for publication November 8, 2013.