Baseline characteristics
Eighty-two patients were enrolled, 45 with and 37 without T2DM. Baseline characteristics of patients are reported in Table 1.
Table 1.Baseline characteristics.
Variable
|
No T2DM (n.37)
|
T2DM (n.45)
|
p-valuea
|
Gender (male), n (%)
|
22 (59.4)
|
25 (55.5)
|
0.448
|
Age (years)
|
66 (58-75)
|
68 (65-71)
|
0.417
|
Weight (Kg)
|
77 (68-91)
|
81 (74-92)
|
0.323
|
BMI (kg/m2)
|
28.2 (25.1-32.3)
|
30.1 (27.4-34.4)
|
0.079
|
Systolic BP (mmHg)
|
140.0 (127.5-158.0)
|
144.0 (135.5-150.5)
|
0.566
|
Diastolic BP (mmHg)
|
80.0 (70.5-87.0)
|
77.0 (70.0-83.0)
|
0.141
|
Hypertension, n (%)
|
33 (89.2)
|
39 (86.6)
|
0.500
|
Fasting plasma glucose (mmol/L)
|
5.2 (4.7-5.6)
|
6.7 (5.7-7.4)
|
<0.001
|
HbA1c (mmol/mol)
|
39.0 (34.0-42.0)
|
51.0 (43.0-56.0)
|
<0.001
|
Total cholesterol (mmol/L)
|
5.0 (4.0-5.6)
|
4.6 (3.9-5.2)
|
0.213
|
LDL cholesterol (mmol/L)
|
2.8 (2.1-3.4)
|
2.4 (2.0-3.0)
|
0.198
|
HDL cholesterol (mmol/L)
|
1.3 (1.1-1.5)
|
1.3 (1.1-1.4)
|
0.588
|
Triglycerides (mmol/L)
|
1.3 (1.0-1.7)
|
1.4 (1.1-1.9)
|
0.386
|
Creatinine (mmol/L)
|
70.4 (61.6-88.0)
|
70.4 (61.6-79.2)
|
0.654
|
eGFR (ml/min)
|
89.3 (72.0-100.0)
|
88.0 (78.0-95.5)
|
0.889
|
hs-C-reactive protein (nmol/L)
|
23.8 (7.6-46.7)
|
29.5 (10.5-63.8)
|
0.404
|
AST (U/L)
|
23.0 (20.5-28.5)
|
25.0 (20.0-31.0)
|
0.712
|
ALT (U/L)
|
27.0 (23.5-32.0)
|
30.0 (21.0-40.5)
|
0.393
|
Uric Acid (mmol/L)
|
351.0 (273.6-422.3)
|
333.1 (276.6-422.3)
|
0.456
|
CVD, n (%)
|
15 (40.5)
|
16 (35.5)
|
0.407
|
Diabetes Duration
|
-
|
6 (2.5-10)
|
-
|
Stable CAD, n (%)
|
1 (2.7)
|
5 (11.1)
|
0.152
|
Previous MI, or revascularization, n (%)
|
6 (16.2)
|
6 (13.6)
|
0.493
|
Previous TIA/stroke, o revascularization, n (%)
|
6 (16.2)
|
2 (4.4)
|
0.079
|
PAD, n (%)
|
0 (0)
|
3 (6.6)
|
0.160
|
Microvascular disease, n (%)
|
0 (0)
|
1 (2.2)
|
0.549
|
Retinopathy, n (%)
|
0 (0)
|
1 (2.2)
|
0.549
|
Chronic renal disease, n (%)
|
1 (2.7)
|
0 (0)
|
0.451
|
Medications
|
|
|
|
Metformin, n (%)
|
1 (2.7)
|
27 (60)
|
<0.001
|
Sulfonylureas, n (%)
|
0 (0)
|
2 (4.4)
|
0.298
|
Glinides, n (%)
|
0 (0)
|
4 (8.8)
|
0.085
|
PPAR-gamma, n (%)
|
0 (0)
|
7 (15)
|
0.012
|
GLP1RA, n (%)
|
0 (0)
|
1 (2.2)
|
0.549
|
DPP-IVi, n (%)
|
0 (0)
|
1 (2.2)
|
0.549
|
Acarbose, n (%)
|
0 (0)
|
1 (2.2)
|
0.549
|
Insulin, n (%)
|
0 (0)
|
0 (0)
|
1.000
|
SGLT2i, n (%)
|
0 (0)
|
0 (0)
|
1.000
|
ACE-I, n (%)
|
11 (29.7)
|
16 (35.5)
|
0.375
|
ARBs, n (%)
|
11 (29.7)
|
10 (22.2)
|
0.301
|
Diuretics, n (%)
|
13 (35.1)
|
12 (26.6)
|
0.278
|
B-block, n (%)
|
19 (51.3)
|
13 (28.8)
|
0.032
|
CCA, n (%)
|
11 (29.7)
|
19 (42.2)
|
0.174
|
Statins, n (%)
|
11 (29.7)
|
25 (55.5)
|
0.016
|
Fibrates, n (%)
|
0 (0)
|
0 (0)
|
1.000
|
Ezetimibe, n (%)
|
0 (0)
|
3 (6.6)
|
0.160
|
Proton Pump Inhibitors, n (%)
|
17 (45.9)
|
18 (40)
|
0.375
|
ASA, n (%)
|
37 (100)
|
45 (100)
|
1.000
|
Abbreviations: BMI=body mass index, BP=blood pressure, CVD=cardiovascular disease, MI= myocardial infarction, TIA=transient ischemic attack, PAD=peripheral artery disease, ACE-I= ACE-inhibitors, ARBs=angiotensin receptor blockers, B-block=beta-blockers, CCA=calcium channel antagonists, ASA=acetylsalicylic acid. Data are median (25th – 75th percentile). aDetermined by Mann-Whitney or x2 test, as appropriate.
Patients were comparable for age (p=0.417), gender (p=0.448), BMI (p=0.079), prevalence of hypertension (p=0.500), CVD (p=0.407) and C-reactive protein (p=0.404).
Patients with T2DM were on statin treatment more frequently than patients without T2DM (p=0.016), while levels of total cholesterol (p=0.213), HDL (p=0.588), and triglycerides (p=0.386) were comparable.
As expected, patients were significantly different for fasting plasma glucose (p<0.001), HbA1c (p<0.001) and diabetes specific treatment such as metformin (p<0.001) and PPAR-gamma agonists (p=0.012). Notably, median HbA1c in T2DM patients was 6.8% (51.0 mmol/mol), reflecting good glycemic control in the group of patients with diabetes included in the study.
Repeatability of the outcome measurement
To verify the intra-subject reproducibility of the main outcome measured, we re-assessed EPC number after 24 hours in 9 out of 82 patients analyzed. We found no significant difference (p=0.573) between the two consecutive days measurements (data not shown).
Effect of T2DM and statin therapy
In patients with T2DM, as compared with patients without T2DM, we observed a trend toward higher levels of plasma PCSK9 (p=0.092) and lower number of CD45neg/CD34bright (p=0.072), CD45neg/CD34bright/CD146neg (p=0.051) and CD45neg/CD34bright/CD309pos (p=0.072) although these trends did not achieve statistical significance. The number of CD45neg/CD34bright/CD146pos was comparable (p=0.754) (Fig. 1).
Thirty-six patients were on statin treatment, 25 with and 11 without T2DM. Patients on statin treatment had a higher prevalence of diabetes, previous myocardial infarction (MI), lower cholesterol and higher transaminase levels. Otherwise, their clinical characteristics were superimposable. Patients on statin treatment, both with and without T2DM, showed higher levels of PCSK9 as compared to their counterparts not on statins (p<0.001 and p=0.036 respectively) (Fig. 2A). Conversely, CD45neg/CD34bright and CD45neg/CD34bright/CD146neg were significantly lower in statin-treated patients than in patients not on statins, only among patients with T2DM (p=0.016 and p=0.040 respectively) (Fig. 2B,C). CD45neg/CD34bright/CD146pos and CD45neg/CD34bright/CD309pos were not affected by statin-treatment either in patients with or without T2DM (Fig. 2D,E).
Interestingly, among subjects not on statin therapy, no difference was detected in either PCSK9 levels or EPCs number between patients with vs. those without T2DM (Fig. 2), while significantly reduced CD45neg/CD34bright (p=0.035) and CD45neg/CD34bright/CD146neg (p=0.032) were observed in patients with diabetes vs. patients without diabetes only among statin-treated subjects (Fig. 2B,C).
Only among patients not on statins, we observed a positive correlation between PCSK9 and total cholesterol (Rho=0.369, p=0.012) and LDL cholesterol (Rho=0.470 p=0.001) (Supplementary Fig.1) but not with HDL or triglycerides (data not shown). Neither PCSK9 nor EPC number were related to fasting plasma glucose or HbA1c (data not shown).
PCSK9 and EPCs
In the whole group, PCSK9 correlated inversely with both CD45neg/CD34bright (Rho=-0.265, p=0.016) and CD45neg/CD34bright/CD146neg (Rho=-0.246, p=0.026) (Fig. 3A,D). In patients with T2DM, PCSK9 correlated inversely with both CD45neg/CD34bright (Rho=-0.409, p=0.006) and CD45neg/CD34bright/CD146neg (Rho=-0.448, p=0.002) (Fig. 3C,F), while such correlation was not observed in patients without T2DM (Fig. 3B,E). None of the other EPC phenotypes explored appeared to be correlated with PCSK9 circulating levels (Supplementary Fig. 2).
Since we have observed that statins affect both PCSK9 levels and CD45neg/CD34bright and CD45neg/CD34bright/CD146neg, we made a sub-analysis as a function of statin treatment. Interestingly, when dividing patients according to statin treatment, we found that the inverse correlations between PCSK9 levels and CD45neg/CD34bright and CD45neg/CD34bright/146neg remained only in patients with diabetes on statin treatment (Rho=-0.454, p=0.022 and Rho=-0.553, p=0.004 respectively) (Fig. 4).
Effect of previous myocardial infarction (MI)
Interestingly, patients with a previous MI (n=12) showed significantly higher levels of plasma PCSK9 (p=0.007), and significantly lower number of CD45neg/CD34bright/CD146neg (p=0.031). Again, in patients with a previous MI, the prevalence of ongoing statin treatment was higher than in subjects without a history of previous MI (75% vs. 38%, p=0.025). The other phenotypes were comparable among patients with and without a previous MI (Supplementary Fig. 3).
Multivariate analyses
Multiple regression analyses performed on all patients, with different EPC phenotypes as the dependent variables, showed that statin treatment was the only independent predictor of the number of CD45neg/CD34bright (β=-0.230; t-value=-2.113; p=0.038) (adjusted R2=0.041, p=0.038). Multiple regression analyses on T2DM patients showed that PCSK9 circulating levels predicted both the number of CD45neg/CD34bright (β=–0.438; t-value, –3.199; p=0.003) (adjusted R2=0.173, p=0.003), and CD45neg/CD34bright/CD146neg (β=–0.458; t-value, –3.376; p=0.002) (adjusted R2=0.191, p=0.002) independently of age, gender, BMI and statin treatment.