Principal findings
The present study of 200 type 2 diabetic patients with microalbuminuria showed an independent predictive value of P-OPG for significant CAD, and although P-OPG correlated with P-NT-proBNP and CCS, its association with CAD remained significant after adjustment for these risk markers. Among patients with significant CAD (n = 70), 23 had >70% coronary artery stenosis at CAG and P-OPG was also independently associated with coronary artery stenosis. Among patients where CAG was performed, 91% of patients with low P-OPG did not have coronary artery stenosis at CAG.
P-OPG as a biomarker for subclinical CAD
Several studies have demonstrated that P-OPG is a strong prognostic cardiovascular risk marker in diabetic and non-diabetic populations [
2‐
5]. For example, we have found that in 283 type 2 diabetic subjects followed for 17 years, high P-OPG predicted all-cause and cardiovascular mortality, independently of known CVD risk markers including levels of albuminuria and P-NT-proBNP[
2]. The latter study was not designed to examine mechanisms behind the increased P-OPG and cardiovascular death. In the current cross-sectional study, we demonstrated that P-OPG was a risk marker for the presence of subclinical CAD, which again may contribute to the association with CVD mortality. OPG expression is increased in the atherosclerotic arterial wall and P-OPG has been suggested to reflect the arterial OPG content and therefore be a surrogate marker of arterial damage[
8]. Along this line, OPG is up-regulated in calcified coronary plaques[
20] and P-OPG has been associated with angiographic disease severity, independent of conventional risk factors in other high risk populations[
9,
10]. One considerably smaller study (n = 40) has previously found that P-OPG was a risk marker for asymptomatic coronary artery stenosis in diabetic patients with at least one additional conventional risk factor[
21]. In the present study, we confirm and extend this observation in a larger and well-defined high risk type 2 diabetic patient group with microalbuminuria. Specifically, the demonstrated association of P-OPG with CAD independent of NT-proBNP or CCS is new and this finding adds significantly to the current knowledge that P-OPG may be considered as an additional marker of atherosclerosis.
Several studies have demonstrated that micro- and macroalbuminura identifies a subgroup of diabetic patients with increased morbidity and mortality from CVD, but a recent study suggested that in the detection of subclinical disease additional markers are in need[
22]. Both NT-proBNP levels and CCS are established markers of CVD mortality with high independent predictive capabilities in numerous populations[
23,
24]. Noteworthy, NT-proBNP levels and CCS in combination have synergistically predictive effect for CVD events in asymptomatic patients without history or signs of CAD followed for 3.9 years[
25]. In contrast, other CVD markers fail to add prognostic value when compared with NT-proBNP and CCS, respectively[
26,
27]. It is therefore interesting in the present study, that P-OPG was independent of NT-proBNP and CCS and this adds to the body of evidence indicating that measurements of P-OPG might be useful as a biomarker for CAD. ROC analyses, however, revealed that the predictive accuracy of OPG alone was low and OPG only slightly improved the covariate model with NT-proBNP and CCS. Of note, the combined model with NT-proBNP and CCS without P-OPG already had an AUC of 0.84 and it is therefore difficult to increase the area further. Along this line, another study reported that P-OPG measurements alone are insufficient as markers of endothelial dysfunction[
28]. Our study suggests that P-OPG could be used in combination with other markers. Previous studies that have used a multi biomarker approach have not included bone metabolism and P-OPG. Bone metabolism with P-OPG measurements may prove important markers of vascular calcification. Actually, CCS and P-OPG in combination could be particular interesting as markers of structural and functional vascular calcification, respectively. In the present study, P-OPG and CCS were related and 80% of patients with low P-OPG (<1
st tertile) also had CCS < 400. Furthermore, in patients with low P-OPG and/or CCS < 400, 83% of patients were also without CAD, however the positive predictive value was low (data not shown). Future studies may show that OPG in combination with CCS, NT-proBNP or other specific plasma markers may provide a strategy to identify patients with subclinical CAD.
If OPG is also a mediator of CVD is not known but OPG may influence plaque morphology and vulnerability through the increased expression of matrix metalloproteinases[
29] Knowledge, however, concerning treatment effects on P-OPG and their possible link to clinical outcome is sparse. An earlier study has showed that low-dose simvastatin treatment for 18 weeks reduced P-OPG in type 2 diabetic patients with microalbuminuria and mild hypercholesterolemia[
30]. Also, a recent study reported that 6 months of pioglitazone decreased P-OPG levels in type 2 diabetic patients[
31]. However, in order to examine an effect on clinical outcomes, substantially longer study periods are of course needed, and a possible therapeutic benefit from lowering P-OPG remains to be shown in larger intervention studies.
Clinical implications
Although it has important prognostic impact to demonstrate subclinical CAD, the therapeutic consequences of diagnosing asymptomatic CAD in diabetic patients remains to be shown. No randomized control trials have as yet examined if myocardial revascularization is beneficial in these asymptomatic patients, but in symptomatic diabetic patients receiving optimal medical therapy there was no difference between the groups receiving revascularization compared to controls[
32].
Accordingly, the recommended primary and secondary CAD treatment remains implementation of intensified medical therapy aimed at reducing the conventional risk factors. The diagnosis of subclinical CAD is likely to encourage the efforts of both the patient and his/her physician to reduce conventional risk factors more effectively, which can diminish CVD morbidity as well as mortality in these patients[
12].
Limitations
As mentioned above, low (first tertile values) P-OPG excluded the presence of >70% coronary artery stenosis in 91% of the patients where CAG was performed according to our risk stratification algorithm. Although this was a high negative predictive value, there is obviously a significant selection bias in play between patients who had a CAG performed in the present study compared to an unselected group of patients with type 2 diabetes and microalbuminuria Furthermore, the positive predictive value of P-OPG for coronary artery stenosis was low. Therefore, the potential use of P-OPG as a rule-out marker needs confirmation in a larger, prospective study.
It is also notable, that we did not examine our 67 low risk patients for significant CAD, since low P-NT-proBNP and/or low CCS is associated with a very good prognosis in these patients[
13]. In this study we therefore assumed in most of our analyses that these 67 patients were without significant CAD. We found, however, that P-OPG was predictive of CAD within the high risk group alone where all patients were investigated for CAD. Other study limitations included the relatively small sample size and that patients with elevated P-creatinine were excluded to allow for examinations with CT and CAG contrast media. Furthermore, all our patients with microalbuminuria were aggressively medically treated, i.e., the value of P-OPG in the diagnosis of asymptomatic CAD in less well-treated patients require further studies.