Baseline characteristics
A flow diagram of the participants in our trial is presented in Supplementary Figure S2. Of the 200 participants screened for eligibility, 120 were excluded. The remaining 80 participants were randomly assigned (1:1) to receive only background lipid-lowering therapy (CLA group) or metformin as an add-on therapy (Met + CLA group). In the CLA and Met + CLA groups, 38 (95.00%) and 33 (82.50%) participants, respectively, completed 1 month of treatment. Six participants in the Met + CLA group discontinued metformin treatment because of diarrhea and nausea or vomiting.
Overall, the average age of the study participants was 58 (interquartile range, 51–66) years, 77% were male, 28% were smokers, 63% had hypertension, the mean BMI was 25.00 kg·m− 2, mean fasting blood glucose concentration was 5.3 mmol·L− 1, and baseline HbA1c was 5.8%. Of the study participants, 100% took statins (atorvastatin 20 mg/d) as the background for lipid-lowering therapy, and 52% were on the ezetimibe regimen. The baseline medium LDL-C, Lp(a), and PCSK9 levels were 76.18 mg·dL− 1, 832.13 mg·dL− 1 (201.30 nmol·L− 1), and 80.54 ng·mL− 1, respectively. Detailed baseline characteristics of the two groups are summarized in Table 1. There was no significant difference in age, gender, smoking, hypertension, BMI, blood glucose, and cardiovascular medication use, including statins, ezetimibe, and other secondary preventions between the two groups (P > 0.05). We also found no marked differences in any of the baseline lipid profiles evaluated using enzymatic or NMR-based methods or in the circulating PCSK9 levels between the two groups (Supplementary Table S1).
Table 1
Baseline characteristics of participants in our cohort.
Variables | CLA group (n = 38) | Met + CLA group (n = 33) | p-Value | |
Age, y | 59.29 ± 11.58 | 56.18 ± 10.41 | 0.241 | |
Men sex, no. (%) | 26(68) | 29(88) | 0.086 | |
Smoking, no. (%) | 10(26) | 10(33) | 0.710 | |
Hypertension, no. (%) | 27(71) | 18(55) | 0.150 | |
BMI* (kg/m2) | 24.92 ± 2.55 | 24.30 ± 4.94 | 0.449 | |
Ezetimibe, no. (%) | 21(55) | 16(48) | 0.546 | |
Other cardiovascular medications — no./total no. (%) | |
Aspirin, P2Y12 inhibitor, or both | 37(97) | 32(97) | 0.919 | |
Beta-blocker | 31(82) | 28(85) | 0.714 | |
ACEi or ARB, aldosterone antagonist, or both | 24(63) | 17(55) | 0.322 | |
Biochemistry Parameters | | |
Fasting glucose (mmol/L) | 5.19 ± 0.89 | 5.49 ± 0.88 | 0.125 | |
HbA1C (%) | 5.80 (5.70,6.00) | 5.75 (5.60,6.00) | 0.695 | |
hs-TnT(pg/mL) | 11.67 (5.32,25.50) | 9.78 (4.54,14.36) | 0.393 | |
hs-CRP(mg/L) | 1.15 (0.62,3.21) | 1.06 (0.34,2.78) | 0.605 | |
ALT(U/L) | 25.25 (16.69,32.37) | 29.50 (16.75,38.10) | 0.253 | |
AST(U/L) | 20.20 (16.00,24.50) | 18.85 (16.20,28.12) | 0.817 | |
eGFR [ml/ (min*1.73 m2)] | 87.29 (77.54, 103.79) | 85.51 (69.04, 103.94) | 0.800 | |
Continuous variables are presented as mean ± SEM or median (interquartile range), and categorical variables are presented as n (%). No significant differences were observed between two groups. Abbreviations: CLA, cholesterol-lowering agents; Met + CLA, metformin plus cholesterol-lowering agents; BMI, body-mass index; ACS, acute coronary syndrome; CCS, chronic coronary syndrome; HbA1c, glycated hemoglobin; hs-CRP, high sensitivity C reactive protein, ALT, alanine aminotransferase, AST, aspartate aminotransferase; ACEi denotes angiotensin- converting enzyme inhibitor, ARB angiotensin-receptor blocker; eGFR, estimated glomerular filtration rate.
* The body-mass index is calculated as the ratio of weight in kilograms to the square of the height in meters.
Metformin as an add-on therapy to CLA resulted in an incremental reduction in the LDL-C level but not that of Lp(a)
After 1 month of exposure, the metformin add-on treatment greatly reduced the LDL-C level [73.47(63.42,90.10) mg/dL vs 54.91(48.72,74.44) mg/dL, (Figure.1A)], allowing 72% and 52% of patients to reach LDL-C < 70 mg/dL and LDL-C < 55 mg/dL, respectively. However, cholesterol-lowering therapy alone (CLA) resulted in a mild reduction in the levels of LDL-C [82.17 (62.16,99.09) mg/dL vs 69.80 (57.23,81.50) mg/dL (Figure.1A) ], resulting in a much lower rate of LDL-C goal achievement (53% for LDL-C < 70 mg/dL and 21% for LDL-C < 55 mg/dL; P < 0.05 for the between-group comparison). Compared to the CLA group, the reduction of LDL-C level was more marked in the Met + CLA group [Δ value: -2.11 (-20.06, 4.79) % vs. -20.81 (-31.60, -1.31) %, P = 0.02 (Figure.1B)]. Unlike the drastic lowering of LDL-C levels in response to metformin add-on treatment, we did not observe the anticipated Lp(a) reduction in the Met + CLA group, which is consistent with previous observations in patients with higher baseline Lp(a) levels [26]. Otherwise, we found that metformin slightly increased the Lp(a) molar (Figure.1C) and mass concentration (Supplementary Figure S3A), whereas the modest elevation of Lp(a) values in the Met + CLA group did not prevail over the CLA group (Figure.1D, Supplementary Figure S3B).
Metformin-mediated incremental reduction in the LDL-C level was attributed to the decrease in the atherogenic LDL particle number
Given that the decrease in LDL-C levels in some conditions may not reflect the true clearance of LDL particles but purely a redistribution of cholesterol in atherogenic LDL subclasses[27], we used the NMR-based method to further confirm the benefit of metformin in improving the LDL atherogenic burden. As expected, in the Met + CLA group, the total LDL particles (LDL-P) were markedly reduced from baseline [853.74(751.04, 997.13) nmol/L vs 763.02(631.20, 903.79) nmol/L, (Figure.2A)], which was primarily due to a significant decrease in the sdLDL-P concentration, but not in the large or medium LDL-P levels (Figure.2B). Moreover, the decline in LDL-P (Figure.2C, P = 0.009) and sdLDL-P (Figure.2D, P = 0.017) levels from baseline in the Met + CLA group were more marked than those in the CLA group, suggesting a favorable effect of metformin on LDL metabolism.
Effect of metformin on TRLs and their composition
Metformin’s ability to improve TRL metabolism in diabetes by alleviating insulin resistance has been reported. We therefore observed the change of TRLs in our cohort after one month of metformin treatment. Except for the downward trend in apoB levels [62.79(56.67, 71.03)vs. 57.65༈48.29, 66.39༉ng·mL− 1, P = 0.08], the concentration of TGs, phospholipids, and cholesterol in TRLs, including VLDL, IDL, and their remnants, were comparable between the two groups (Table 2).
Table 2
Effects of 1-month of metformin treatment on triglyceride-rich lipoproteins (TRLs) and their components.
Variables | CLA group (n = 38) | Met + CLA group (n = 33) | Change in the Met + CLA group (%) | p-Value |
Enzymatic method |
TG (mg/dL) | 105.84* (78.38,158.76) | 89.46† (64.66,143.93) | -12.8 (-36.62,28.86) | 0.572 |
TRL-C(mg/dL) | 18.56 (14.31,22.43) | 15.85 (13.73,20.11) | 0.00 (-21.92,37.12) | 0.628 |
NMR-based method |
VLDL-P (nmol/L) | 132.80‡ (102.84,206.41) | 117.38† (89.22,190.85) | -10.72 (-32.07,6.10) | 0.854 |
IDL-P (nmol/L) | 52.78 (43.82,73.81) | 59.72 (46.21,71.07) | 8.65 (-22.49,50.55) | 0.226 |
VLDL-C (mg/dL) | 15.45‡ (10.52,25.44) | 11.03† (8.75,22.14) | -15.81 (-44.28,11.43) | 0.945 |
IDL-C (mg/dL) | 7.95 (5.41,11.89) | 7.73 (5.80,10.62) | 2.38 (-31.46,47.35) | 0.786 |
VLDL-TG (mg/dL) | 67.83* (49.67,107.83) | 51.99§ (42.32,91.91) | -22.93 (-41.58,16.99) | 0.670 |
IDL-TG (mg/dL) | 6.06* (2.23,13.68) | 2.34§ (0.91,9.63) | -60.09 (-81.70, -16.67) | 0.102 |
VLDL-Pho (mg/dL) | 17.70* (13.26,27.83) | 14.54† (11.36,25.03) | -17.54 (-36.70,6.54) | 0.782 |
IDL-Pho (mg/dL) | 4.38 (2.68,7.28) | 3.42 (2.25,5.86) | -16.26 (-48.15,39.81) | 0.383 |
RC (mg/dL) | 15.69 (12.51,24.58) | 14.96 (11.30,20.65) | 12.19 (-30.87,35.59) | 0.426 |
apoB (mg/dL) | 62.79 (56.67,71.03) | 57.65§ (48.29,66.39) | -6.89 (-24.32,1.57) | 0.080 |
VLDL-apoB (mg/dL) | 7.3‡ (5.66,11.36) | 6.46† (4.91,10.50) | -10.67 (-32.00,6.17) | 0.872 |
IDL-apoB (mg/dL) | 2.91 (2.41,4.06) | 3.28 (2.55,3.91) | 8.95 (-22.47,50.42) | 0.213 |
All data are presented as median with an interquartile range and % change. apoB indicates apolipoprotein B; IDL, intermediate-density lipoprotein; LDL, low-density lipoprotein; LDL-C, low-density lipoprotein cholesterol; TG, triglycerides; VLDL, very low-density lipoprotein; and Pho, phospholipid. ‡, p < 0.05 in CLA group compared with baseline; *, p < 0.01 in CLA group compared with baseline; †, p < 0.05 in Met + CLA group compared with baseline; §, p < 0.01 in Met + CLA group compared with baseline.
Metformin significantly decreased circulating PCSK9 levels by inhibiting hepatic expression of PCSK9
Our previous study demonstrated that metformin could downregulate hepatic PCSK9 mRNA and protein expression via the intracellular-glucose sensor ChREBP pathway, both in cultured cells and animal models [23]. Supporting our aforementioned mechanistic research, the circulating PCSK9 concentration was markedly reduced from baseline by 15% in the metformin-allocated group [80.54(64.30, 116.34)vs. 63.91༈54.58, 77.66༉ng·mL− 1, Δ value − 15.03 (-37.10, -1.38), P = 0.000]. In contrast, the serum PCSK9 level in the CLA group did not show a statistically significant difference from baseline (Figure.3A). Notably, metformin treatment did not cause hypoglycemia in CAD patients without diabetes, while statin +/- ezetimibe treatment exhibited a moderate elevation in blood glucose concentration.
Furthermore, although there was no significant blood glucose lowering in the Met + CLA group, we observed that after metformin treatment, the decrease in the serum PCSK9 level was highly related to the levels of blood glucose and ketone bodies, which were both potent regulators of ChREBP functions (Supplementary Figure S4, Supplementary Table S2). Moreover, the reduction in the serum PCSK9 level was strongly associated with the decline in LDL-P (r = 0.44, P = 0.01, Figure.3B) and sdLDL-P (r = 0.42, P = 0.02, Figure.3C) levels only in the Met + CLA group but not in the CLA group.
To identify the effect of metformin on statin-induced PCSK9 expression, we incubated HepG2 hepatocytes with simvastatin or metformin plus simvastatin for 24h to evaluate the expression of PCSK9. Simvastatin treatment increased both mRNA and protein levels of PCSK9 compared to those in the control group; however, increased PCSK9 expression was inhibited by metformin (Figure.4).