Predictors and correlates of systolic blood pressure reduction with liraglutide treatment in patients with type 2 diabetes

Abstract

Liraglutide is associated with blood pressure reduction in patients with type 2 diabetes. However, it is not known whether this blood pressure reduction can be predicted prior to treatment initiation, and to what extent it correlates with weight loss and with improved glycemic control during follow‐up. We analyzed data from a double‐blind, placebo‐controlled trial, in which 124 insulin‐treated patients with type 2 diabetes were randomized to liraglutide or placebo. We evaluated various baseline variables as potential predictors of systolic blood pressure (SBP) reduction, and evaluated whether changes in SBP correlated with weight loss and with improved glycemic control. A greater reduction in SBP among liraglutide‐treated patients was predicted by higher baseline values of SBP (P < 0.0001) and diastolic blood pressure (P = 0.012), and by lower baseline values of mean glucose measured by continuous glucose monitoring (CGM; P = 0.044), and serum fasting C‐peptide (P = 0.015). The regression coefficients differed significantly between the liraglutide group and the placebo group only for diastolic blood pressure (P = 0.037) and mean CGM (P = 0.021). During the trial period, SBP reduction correlated directly with change in body weight and BMI, but not with change in HbA1c. We conclude that patients with lower mean CGM values at baseline responded to liraglutide with a larger reduction in SBP, and that improved HbA1c during follow‐up was not associated with reductions of SBP. Our data suggest that some patients with type 2 diabetes may benefit from liraglutide in terms of weight and SBP reduction.

1. INTRODUCTION

Elevated blood pressure contributes considerably to the increased risk of cardiovascular disease associated with diabetes,1 and it is well established that treatment with blood pressure‐lowering medications reduces the risk for both macro‐ and microvascular diabetes complications.2 Despite this, blood pressure frequently remains sub‐optimally controlled in patients with diabetes.3, 4 Therefore, finding better means to lower blood pressure is an important issue for improving the outcome of patients with diabetes. Interestingly, agonists to the glucagon‐like peptide 1 (GLP‐1) receptor, which are used primarily to improve glycemic control in patients with type 2 diabetes, have been associated with significant reductions in systolic blood pressure (SBP). This was first demonstrated with exenatide in patients with type 2 diabetes in a small study5 and has thereafter been confirmed with various GLP‐1 receptor agonists in larger randomized controlled trials in patients with type 2 diabetes6, 7, 8, 9 and also with liraglutide in patients with type 1 diabetes.10 It is not known, however, whether the blood pressure‐lowering effect of GLP‐1 receptor agonists can be predicted by factors determined prior to treatment initiation. It is also not known whether GLP‐1 receptor agonist‐induced blood pressure reduction is associated with weight loss or with improved glycemic control. The glucose‐lowering efficacy of the GLP‐1 receptor agonist liraglutide, added to patients already treated with multiple daily insulin injections, was recently demonstrated in the randomized, placebo‐controlled MDI Liraglutide trial.11 Further analyses from this study have revealed that whereas some patients responded predominantly with glycemic improvement, other patients responded with weight reduction, suggesting inter‐individual heterogeneity in terms of which risk factors that are affected by liraglutide.12 The primary aim of the present study (MDI Liraglutide study 4) was to identify potential predictors of SBP reduction by liraglutide in patients with type 2 diabetes who were already treated with multiple daily insulin injections. We also sought to explore associations between SBP reduction and reductions in body weight, body mass index (BMI), and hemoglobin A1c (HbA1c) associated with liraglutide treatment in these patients.

2. METHODS AND PATIENTS

2.1. Patients and study procedure

The study design and the main outcomes of the trial have been described in detail.11, 13 The trial was approved by the ethics committee of the University of Gothenburg, Gothenburg, Sweden (diary No 596‐12). It was registered in the EudraCT database before study start (EudraCT No 2012‐001941‐42). The study was a double‐blind, placebo‐controlled parallel group trial, in which 124 patients were randomized to treatment with either liraglutide (1.8 mg/d) or placebo for 24 weeks. In brief, inclusion criteria were type 2 diabetes treated with multiple daily insulin injections with or without metformin (patients with pre‐mixed insulin were excluded), HbA1c ≥ 58 mmol/mol and ≤102 mmol/mol, BMI 27.5‐45.0 kg/m² and fasting C peptide level ≥0.1 nmol/L. Masked continuous glucose monitoring (CGM) was performed at baseline and at week 12 and during one of the two last weeks of the trial, using a subcutaneous glucose sensor which measured glucose values continuously during 1 week (Dexcom G4 PLATINUM system, San Diego, CA). Blood pressure was measured at baseline and at weeks 6, 12, 18, and 24, together with HbA1c, body weight, and BMI. Blood pressure measurements were performed after at least 5 minutes rest, in the seated position, using appropriately sized cuffs. The timing of the blood pressure measurements in relation to the liraglutide injections was not standardized in the study protocol, but at baseline, week 12, and week 24, blood pressure measurements were typically performed in the morning when patients came fasting to their study center for blood samples to be taken. Data concerning insulin doses, hypoglycemic events and adverse events were recorded on each occasion. Waist and hip circumference and abdominal sagittal diameter were measured at baseline and at weeks 12 and 24. The primary end‐point of the main study was a change in HbA1c between baseline and week 24, and the study was powered to detect a difference in HbA1c of 7 mmol/mol between liraglutide and placebo.

2.2. Statistics

For descriptive purposes, data are presented as mean, standard deviation (SD), median, and range (min; max) for continuous variables and as n (%) for categorical variables. Baseline comparisons between groups were performed with Fisher’s non‐parametric permutation test for continuous variables and with Fisher’s exact test for dichotomous variables. Comparisons between treatment groups with respect to change in blood pressure were performed with ANCOVA, adjusting for blood pressure at baseline. Prediction analyses of changes in SBP were performed using linear regression with various baseline variables as predictors, analyzed one at a time, and with analyses performed separately in the liraglutide group and in the placebo group. Statistical tests for the effect of the predictors were performed in the liraglutide group only. For those predictors that were statistically significant in the liraglutide group at the 5% level, we also evaluated whether their predictive ability were significantly different in the liraglutide group than in the placebo group. This was tested by interaction analysis, in order to understand if the effects of the predictors were indeed related to the use of liraglutide, or to other study related causes. Finally, post hoc multivariable analyses were performed, including only predictors significant at the 5% level both in the liraglutide group and in the subsequent interaction analyses. The following baseline variables were chosen a priori to be evaluated as potential predictors: age, sex, weight, BMI, waist circumference, waist‐hip ratio, abdominal sagittal diameter, SBP, diastolic blood pressure (DBP), HbA1c, mean glucose during CGM, total daily insulin dose, diabetes duration, and concentrations of serum fasting C‐peptide. We also performed similar univariate analyses for all commonly used antihypertensive drug classes. We performed analyses of correlations, by means of calculations of the Pearson correlation coefficients, between change in SBP from baseline to week 6, 12, and 24 and changes in HbA1c, weight, BMI, and total daily insulin dose, and between change in SBP from baseline to week 12 and 24 and changes in waist circumference, abdominal sagittal diameter and hip circumference, since all of these parameters are commonly used in clinical practice to evaluate the efficacy of glucose‐lowering drugs. Correlation analyses were also performed for SBP at baseline and baseline values of these variables. We also evaluated the proportion of patients who achieved clinically significant reductions in SBP (>5 mm Hg), HbA1c (>10 mmol/mol) or weight loss (>3%). All statistical tests were performed at 5% significance level. All analyses were performed with SAS version 9.4 (Cary, North Carolina, USA).

3. RESULTS

3.1. Baseline characteristics

Of the 124 participants in the trial, there were 122 patients with at least one valid follow‐up measurement, which were included in the present analysis. Their baseline characteristics are described in Table 1. The median number of antihypertensive medications used did not differ between groups (two in both groups, P = 0.90). Most patients were of European ethnicity, only two patients (both in the liraglutide group) were of Hispanic/Latino ethnicity.

Table 1.

Baseline characteristics

Variable	Liraglutide (n = 63)	Placebo (n = 59)	P
Age, y
Mean (SD)	63.8 (8.2)	63.6 (7.7)	0.88
Median (range)	66.3 (44.1; 78.0)	65.0 (38.9; 77.3)
Sex, n (%)
Male	40 (63.5)	39 (66.1)	0.91
Female	23 (36.5)	20 (33.9)
Systolic blood pressure, mm Hg
Mean (SD)	137.9 (16.8)	133.7 (13.7)	0.14
Median (range)	139.0 (101.0; 180.0)	134.0 (104.0; 157.0)
Diastolic blood pressure, mm Hg
Mean (SD)	73.5 (12.7)	74.9 (8.5)	0.48
Median (range)	74.0 (45.0; 103.0)	76.0 (54.0; 97.0)
BMI, kg/m2
Mean (SD)	33.7 (4.3)	33.5 (4.0)	0.75
Median (range)	33.3 (27.3; 44.0)	33.5 (27.7; 43.0)
Weight, kg
Mean (SD)	98.8 (14.1)	99.8 (14.8)	0.70
Median (range)	100.0 (69.0; 134.9)	96.0 (72.5; 139.2)
Abdominal sagittal diameter, cm
Mean (SD)	27.9 (3.5)	27.8 (3.5)	0.78
Median (range)	27.5 (20.5; 36.9)	27.2 (22.0; 36.7)
Waist:hip ratio
Mean (SD)	1.03 (0.07)	1.04 (0.06)	0.54
Median (range)	1.04 (0.82; 1.16)	1.04 (0.90; 1.22)
Fasting plasma glucose, mmol/L
Mean (SD)	9.96 (3.17)	9.41 (2.55)	0.30
Median (range)	9.40 (4.20; 17.90)	9.40 (2.50; 19.60)
HbA1c (IFCC), mmol/mol
Mean (SD)	74.6 (10.8)	74.4 (12.0)	0.92
Median (range)	73.0 (53.0; 103.0)	73.0 (54.0; 101.0)
HbA1c (NGSP), %
Mean (SD)	8.98 (0.99)	8.96 (1.10)	0.91
Median (range)	8.83 (7.00; 11.58)	8.83 (7.09; 11.39)
Average CGM, mmol/L
Mean (SD)	10.9 (2.3)	10.7 (2.2)	0.56
Median (range)	10.5 (5.7; 16.6)	10.0 (6.9; 16.7)
Fasting total cholesterol, mmol/L
Mean (SD)	4.18 (0.92)	4.24 (1.00)	0.73
Median (range)	4.10 (2.70; 7.90)	4.10 (2.40; 6.80)
Fasting LDL cholesterol, mmol/L
Mean (SD)	2.22 (0.79)	2.28 (0.96)	0.70
Median (range)	2.10 (0.20; 4.40)	2.30 (0.50; 4.80)
Fasting HDL cholesterol, mmol/L
Mean (SD)	1.12 (0.23)	1.07 (0.32)	0.39
Median (range)	1.10 (0.70; 1.80)	1.00 (0.60; 2.80)
Fasting triglycerides, mmol/L
Mean (SD)	1.87 (1.11)	2.15 (1.59)	0.27
Median (range)	1.60 (0.59; 6.50)	1.65 (0.56; 9.60)
Fasting C‐peptide, nmol/L
Mean (SD)	0.651 (0.477)	0.727 (0.494)	0.39
Median (range)	0.560 (0.090; 3.100)	0.620 (0.100; 2.600)
Thiazide diuretics, n (%)	7 (11.5)	4 (6.8)	0.57
Loop diuretics, n (%)	10 (16.4)	13 (22.0)	0.58
Beta blockers, n (%)	22 (36.1)	27 (45.8)	0.37
Calcium channel blockers, n (%)	25 (41.0)	14 (23.7)	0.07
ACE inhibitors or ARB, n (%)	50 (82.0)	42 (71.2)	0.24
Number of antihypertensive medications, n (%)
0	4 (6.6)	8 (13.6)	0.90
1	24 (39.3)	19 (32.2)
2	15 (24.6)	20 (33.9)
3	12 (19.7)	8 (13.6)
4	6 (9.8)	3 (5.1)
5	0 (0.0)	1 (1.7)
Metformin, n (%)	43 (68.3)	43 (72.9)	0.72
Total daily basal insulin dose, units
Mean (SD)	57.2 (25.9)	59.3 (26.4)	0.66
Median (range)	54.0 (12.0; 130.0)	60.0 (18.0; 130.0)
Total daily meal insulin dose, units
Mean (SD)	48.1 (25.6)	46.3 (26.6)	0.70
Median (range)	40.0 (12.0; 114.0)	40.0 (8.0; 165.0)
Total number of insulin injections
Mean (SD)	4.46 (0.88)	4.42 (0.62)	0.89
Median (range)	4.00 (3.00; 9.00)	4.00 (3.00; 6.00)
Current Smoker	8 (12.7)	7 (11.9)	1.00
Diabetes duration, y
Mean (SD)	17.3 (7.7)	17.0 (8.2)	0.88
Median (range)	16.0 (4.0; 40.0)	16.0 (2.0; 35.0)
Past complications, n (%)
MI	6 (9.5)	10 (16.9)	0.34
Stroke	1 (1.6)	0 (0.0)	1.00
PCI	5 (7.9)	8 (13.6)	0.48
Bypass Surgery	5 (7.9)	7 (11.9)	0.67
Retinopathy (treated with laser)	9 (14.3)	14 (23.7)	0.27
Past foot (or leg) ulcer	3 (4.8)	4 (6.8)	0.93

ACE, Angiotensin Converting Enzyme; ARB, Angiotensin Receptor Blocker; CGM, Continuous Glucose Monitoring; HbA1c, Hemoglobin A1c; HDL, High Density Lipoprotein; IFCC, International Federation of Clinical Chemistry; LDL, Low Density Lipoprotein; MI, Myocardial Infarction; NGSP, National Glycohemoglobin Standardization Program; PCI, Percutaneous Coronary Intervention; SD, Standard Deviation.

For categorical variables, n (%) is presented.

For continuous variables, mean (SD) and median (min; max) are presented.

Data missing for 1 patient (HDL cholesterol, triglycerides, total cholesterol and abdominal sagittal diameter); 3 patients (CGM mean); 5 patients (waist:hip ratio) and 8 patients (LDL cholesterol), respectively.

3.2. Overall changes in blood pressure variables

A significant reduction of SBP was observed in the liraglutide group already after 6 weeks of treatment (Figure 1). At week 24, SBP was significantly reduced in the liraglutide group (mean change from baseline was −5.69 mm Hg; 95% CI −9.11 to −2.26 mm Hg) but not in the placebo group (mean change from baseline was 1.98 mm Hg; 95% CI −1.52 to 5.48 mm Hg). The change was significantly greater in the liraglutide group than in the placebo group (baseline adjusted mean difference between groups was −5.47 mm Hg; 95% CI −9.85 to −1.10 mm Hg, P = 0.015). During the course of the trial, DBP did not change significantly (Figure 1) neither in the liraglutide group (mean change from baseline to week 24 was 0.70 mm Hg; 95% CI −1.94 to 3.35 mm Hg) nor in the placebo group (mean change from baseline to week 24 was 0.22 mm Hg; 95% CI −2.58 to 3.03 mm Hg). At week 24, there was no statistically significant difference between the groups for change in DBP (baseline adjusted mean difference between groups was 0.25 mm Hg; 95% CI −3.04 to 3.54 mm Hg, P = 0.88).

Figure 1.

Changes in systolic blood pressure (A, left panel) and diastolic blood pressure (B, right panel) during 24 wk of treatment with liraglutide or placebo. At week 24, P values for baseline‐adjusted mean differences between groups were 0.015 (systolic blood pressure change) and 0.88 (diastolic blood pressure change)

3.3. Baseline variables predicting systolic blood pressure reduction

Regression coefficients for all evaluated potential predictors of SBP reduction, as well as regression coefficients for individual antihypertensive drug classes are presented in Table S1. We did not detect any influence of previous antihypertensive medications on the SBP lowering effect of liraglutide (Table S1). The following four baseline variables were statistically significant predictors of SBP reduction in the liraglutide group: SBP, DBP, serum fasting C‐peptide, and mean CGM (Table S1). In the liraglutide group, a 10 mm Hg higher baseline value of SBP was associated with a 4.51 mm Hg larger reduction of SBP (P < 0.0001), a 10 mm Hg higher baseline value of DBP was associated with a 3.29 mm Hg larger reduction of SBP (P = 0.012), a one mmol/L lower baseline mean CGM value was associated with a 1.51 mm Hg larger reduction of SBP (P = 0.044), and a 0.2 nmol/L lower baseline value of fasting C‐peptide was associated with a 2.29 mm Hg larger reduction of SBP (P = 0.015). In Figure 2, the regression lines for change in SBP from baseline to week 24 are plotted against baseline values of these four predictors. Thus, the slopes of the regression lines in Figure 2 correspond to the regression coefficients presented in Table S1, and differed significantly between the liraglutide group and the placebo group for the predictors baseline DBP and mean CGM (tests for treatment with predictor interaction: P = 0.037 and P = 0.021, respectively, Table S1), but not for the predictors baseline SBP and fasting C‐peptide (tests for treatment with predictor interaction: P = 0.55 and P = 0.14, respectively, Table S1). This indicates that only DBP and mean CGM differed significantly between the liraglutide group and the placebo group in terms of strength as predictors for SBP reduction, with DBP being a stronger predictor in the placebo group and mean CGM being a stronger predictor in the liraglutide group.

Figure 2.

Regression lines with 95% confidence intervals for change in systolic blood pressure from baseline to week 24, vs baseline values of diastolic blood pressure (A, upper left panel), mean continuous glucose monitoring (CGM) (B, upper right panel), systolic blood pressure (C, lower left panel), and fasting C‐peptide (D, lower right panel). The slopes of the regression lines correspond to the regression coefficients in Table S1. The P values for tests for treatment with predictor interactions were as follows: 0.037 (diastolic blood pressure); 0.021 (CGM); 0.55 (systolic blood pressure); and 0.14 (fasting C‐peptide)

3.4. Post hoc multivariable analysis of predictors for lower systolic blood pressure

Since DBP and mean CGM were predictors of SBP reduction they were tested in a multivariable analysis for prediction of change in SBP. Diastolic blood pressure remained a significant predictor of SBP reduction when adjusting for mean CGM, both within the liraglutide group (P = 0.011) and compared with placebo (test for treatment with predictor interaction: P = 0.046). Likewise, mean CGM remained a significant predictor of SBP reduction when adjusted for DBP, both within the liraglutide group (P = 0.021) and compared with placebo (test for treatment with predictor interaction: P = 0.010).

3.5. Patients improving both in blood pressure, weight and HbA1c

The numbers and proportions of patients experiencing decreased SBP, weight, and HbA1c after 24 weeks of treatment are shown in Figure 3. A significantly larger proportion of patients in the liraglutide group than in the placebo group experienced a reduction of SBP by more than 5 mm Hg (52.4% vs 25.4%, P = 0.004), a reduction of body weight by more than 3% (55.6% vs 5.1%, P < 0.0001) or a reduction of HbA1c by more than 10 mmol/mol (76.2% vs 27.1%, P < 0.0001). Likewise, a significantly larger proportion of patients in the liraglutide group than in the placebo group experienced simultaneous reductions in all three measures (30.2% vs 1.7%, P < 0.0001). Out of the 33 patients in the liraglutide group whose SBP decreased more than 5 mm Hg, there were six patients (18.2%) who did not experience a simultaneous decrease of HbA1c by more than 10 mmol/mol and 11 patients (33.3%) who did not experience a simultaneous decrease in body weight by more than 3%. The mean reduction in SBP among liraglutide‐treated patients with HbA1c reduction greater than 10 mmol/mol was −6.1 mm Hg (95% CI −10.1 to −2.1), which was not significantly different from the reduction of SBP observed in liraglutide‐treated patients with smaller HbA1c reduction (−4.5 mm Hg; 95% CI −11.4 to 2.5), 95% CI for between‐group difference −9.6 to 6.4 mm Hg, P = 0.69. The mean reduction in SBP in liraglutide‐treated patients with weight reduction greater than 3% was −9.2 mm Hg (95% CI −13.7 to −4.8), which was significantly greater than the reduction in SBP in liraglutide‐treated patients with smaller weight reduction (−1.2 mm Hg; 95% CI −6.2 to 3.7), 95% CI for between‐group difference −14.7 to −1.4 mm Hg, P = 0.019.

Figure 3.

Numbers and proportions of liraglutide‐ and placebo‐treated patients with systolic blood pressure reduction >5 mm Hg, weight loss >3%, and HbA1c decrease >10 mmol/mol, after 24 weeks of treatment. **P < 0.01; ****P < 0.0001

3.6. Correlation analyses

Correlation coefficients between SBP and HbA1c, weight, BMI, waist circumference, abdominal sagittal diameter, hip circumference, total daily meal, and basal insulin dose at baseline and for changes in these variables from baseline to different time points during the trial period, are shown in Figure 4 for the patients in the liraglutide group. No significant correlations were found at baseline. Change in SBP correlated significantly with change in weight at week 12 (r = 0.28, P = 0.029) and at week 24 (r = 0.26, P = 0.041), and with change in BMI at week 12 (r = 0.28, P = 0.030) and at week 24 (r = 0.30, P = 0.019).

Figure 4.

Correlations between systolic blood pressure, HbA1c, weight, body mass index (BMI), waist circumference (WC), abdominal sagittal diameter (ASD), hip circumference (HC), and total daily meal and basal insulin dose (TDI) at baseline and for changes from baseline to week 6, 12, and 24, in patients in the liraglutide group. No follow‐up measurements of WC, ASD, and HC were made at week 6. Error bars are 95% confidence intervals for the correlation coefficients

4. DISCUSSION

In this randomized double‐blind placebo‐controlled clinical trial, liraglutide was associated with a significant reduction in SBP compared with placebo, when added to patients with type 2 diabetes who were already treated with multiple daily insulin injections. The main finding of the present analysis was that a larger reduction of SBP was predicted by higher baseline values of DBP and by lower baseline values of mean CGM. Furthermore, larger reductions in SBP during follow‐up was associated with weight loss and with BMI reduction, but not with HbA1c reduction. Finally, we identified a sub‐group of patients who responded to liraglutide with a clinically significant reduction in systolic blood pressure without concomitant reductions of HbA1c or body weight. To the best of our knowledge, the present study represents the first attempt to identify predictors of a favorable blood pressure response to liraglutide.

4.1. Clinical considerations

High blood pressure is an important modifiable risk factor for cardiovascular disease in patients with type 2 diabetes. A reduction of SBP by about 5 mm Hg, as was observed with liraglutide in this placebo‐controlled trial, is larger than what has been reported in clinical trials of GLP‐1 receptor agonists previously,6, 7, 8, 9 and is likely to be of clinical significance. A similar degree of SBP reduction has been estimated to lower the risk for stroke by 13% in patients with diabetes.14 Likewise, a reduction of HbA1c by a magnitude similar to that observed with liraglutide in the present trial has been associated with a 15% lower risk for myocardial infarction.15 In our study, a significantly larger proportion of patients in the liraglutide group than in the placebo group responded with simultaneous reductions of both HbA1c and SBP, and with decreased body weight. This is an important observation, since interventions that target multiple risk factors have been particularly successful in patients with type 2 diabetes.16 However, one noteworthy finding of the present analysis was that not all patients who experienced a clinically relevant SBP reduction experienced a simultaneous reduction of HbA1c and body weight. Indeed, we found that in liraglutide‐treated patients with a marked reduction in SBP, almost one out of five (6/33 or 18.2%) did not lower their HbA1c markedly, and one out of three (11/33 or 33.3%) did not decrease their body weight markedly. This indicates that some patients may benefit from liraglutide from a blood pressure perspective, without exhibiting improvements of traditional metabolic risk factors. This novel finding may have important clinical implications. For instance, it has been suggested that GLP‐1 receptor agonists should be discontinued in patients not experiencing a reduction in HbA1c >10 mmol/mol and a simultaneous reduction of body weight ≥3% after 24 weeks of treatment.17 Our data suggest that in future guidelines, as well as in the evaluation of individual patients, the SBP response should also be taken into account when the efficacy of liraglutide is evaluated.

4.2. Possible pharmacological mechanisms

Pharmacological mechanisms by which GLP‐1 receptor agonists lower blood pressure remain incompletely understood. Weight loss may be one factor that explains the blood pressure reduction observed in patients treated with liraglutide. Weight reducing diets typically lower SBP by about 4.5 mm Hg compared with control diets in randomized clinical trials.18 Weight loss after Roux‐en‐Y gastric bypass surgery has been associated with increased plasma concentrations of GLP‐1,19 with increased plasma concentrations of atrial natriuretic peptide (ANP) and with decreased plasma concentrations of angiotensinogen, renin and angiotensin II20 as well as with improved blood pressure control.21 A weak but statistically significant association between weight loss and SBP reduction has been reported previously in a pooled analysis of randomized liraglutide trials.22 On the other hand, in a meta‐analysis of trials of various GLP‐1 receptor agonists, weight loss did not predict SBP reduction,23 but less than half of the patients in that meta‐analysis used liraglutide, so different drugs may exert different actions in this respect. We found significant correlations between reductions in SBP and reductions in body weight and BMI. This might offer a plausible explanation to why lower baseline mean CGM values predicted a more advantageous blood pressure response to liraglutide: patients with higher mean CGM values were more likely to experience liraglutide‐induced glycemic improvement, which would decrease glucosuria and thereby attenuate weight loss. This line of reasoning is also supported by our previous finding that higher baseline HbA1c predicted a less marked weight loss by liraglutide.12 However, in the present analysis, we found that SBP was significantly reduced already after 6 weeks, whereas the associations between weight loss and SBP reduction and between BMI reduction and SBP reduction did not reach statistical significance until week 12. Thus, other mechanisms than weight loss may also be of importance. It has been proposed, for instance, that increased natriuresis may contribute to the blood pressure‐lowering effect of liraglutide. In one study, liraglutide treatment was found to increase urinary sodium excretion,24 and in another study, liraglutide was found to increase the levels of natriuretic peptides.25 In insulin‐treated patients with type 2 diabetes, treatment with the GLP‐1 receptor agonist exenatide has also been shown to increase plasma concentrations of the vasodilators cyclic guanyl monophosphate (cGMP) and cyclic adenyl monophosphate (cAMP), and to reduce plasma concentrations of angiotensinogen, renin and angiotensin II.26 Under physiological conditions, insulin increases the expression and induction of endothelial nitric oxide synthase (eNOS), but in insulin‐resistant conditions such as type 2 diabetes and obesity, this vasodilating effect of insulin is attenuated.27 GLP‐1 receptors are expressed in endothelial cells,28 and it has been suggested that GLP‐1 receptor agonists may lower blood pressure by improving endothelial function. Improved endothelial function has been demonstrated with exenatide both acutely29 and after prolonged treatment,30 whereas no effect on endothelial function was observed with liraglutide.31 In the present study, we did not measure neither urinary sodium excretion, plasma natriuretic peptides, cGMP, cAMP, angiotensinogen, renin, angiotensin II nor endothelial function, so we can only speculate on the mechanisms that might explain the blood pressure‐lowering effect of liraglutide. It seems likely that multiple pathways are involved, and that they may differ between individuals. Since associations do not necessarily imply causality, further mechanistic studies seem warranted.

4.3. Study strengths and limitations

The randomized placebo‐controlled trial design allowed us to make comparisons with a placebo group, which we consider a major strength in this study of predictive variables. Reduction of SBP with liraglutide treatment was larger in patients with, for instance, lower compared with higher baseline values of mean CGM, but also in patients with, for instance, higher compared with lower values of SBP at baseline. Interaction analyses with the placebo group as comparator allowed us to demonstrate that a lower baseline mean CGM value was indeed a stronger predictor of SBP reduction in the liraglutide group than in the placebo group, whereas higher baseline values of SBP predicted larger SBP reductions equally well in the liraglutide group and in the placebo group. We conclude that the role of high baseline SBP as a predictor for SBP reduction is explained by “regression to the mean,” a phenomenon which occurs when a single measurement at baseline deviates from the individual’s long‐term average value due to non‐systematic measurement errors.32 From a methodological point of view, our findings highlight the importance of comparing apparent predictors of treatment‐induced effects with observations from a placebo group, since regression to the mean will be observed both in actively treated patients and in placebo‐treated patients.33 We consider the relatively small number of patients as the most important study limitation, since this precluded us from performing subgroup analyses, and also calls for a cautious interpretation of our data, particularly concerning possible interactions between previous antihypertensive medications and the study intervention. In two recent studies, ambulatory blood pressure levels were not significantly reduced by liraglutide treatment compared with placebo34 and glimepiride,35 respectively, but both studies were small so the apparent lack of effect may simply have been due to low statistical power. Nonetheless, it should be acknowledged that the fact that we did not evaluate ambulatory blood pressure levels in the present study could be considered a limitation. Furthermore, we did not collect data on heart rate, which has been reported to increase with GLP‐1 receptor agonist treatment in several other randomized clinical trials.6, 7, 8, 9 Other study limitations include that the follow‐up period was restricted to 24 weeks, and that our results may not be extrapolated to patients with type 2 diabetes who do not use multiple daily insulin injections, or to patients of different ethnic backgrounds. Since this is an exploratory analysis of our trial data, it would be valuable to confirm (or refute) our findings in a larger trial.

5. CONCLUSIONS

To summarize, we found (a) that a larger SBP reduction by liraglutide was predicted by higher baseline values of DBP and by lower baseline mean CGM values, (b) that a larger reduction in SBP during follow‐up was associated with weight loss and with BMI reduction, but not with HbA1c reduction, and (c) that some patients responded to liraglutide with SBP reduction without experiencing clinically significant improvements of HbA1c or body weight. Together with the recent analysis of predictors of weight reduction by liraglutide,12 the present study suggests that patients with better glycemic control may respond to liraglutide with greater weight and blood pressure reductions, whereas those with a higher HbA1c are more likely to respond with improved glycemic control. Thus, our data support an individualized treatment evaluation of liraglutide in patients with type 2 diabetes, in which not only changes in metabolic parameters, but also changes in SBP and weight should be taken into account.

CONFLICT OF INTEREST

MW has served on advisory boards and/or lectured for MSD, Lilly, Novo Nordisk, and Sanofi and has organized a professional regional meeting sponsored by Lilly, Rubin Medical, Sanofi, Novartis, and Novo Nordisk. SD’s and ML’s institution received grants from Novo Nordisk during the conduct of the study. IH has been a consultant for Abbott Diabetes Care, Adocia, Intarcia, Roche, Bigfoot, and Valeritas. JT has received grants from Bayer Pharma, Boehringer Ingelheim, Merck Serono, and MSD outside the submitted work, and has acted as a consultant, advisory board member or speaker for Merck Serono, Orion Pharma, Renova, and MSD, and is a share owner in Orion Pharma. OT has acted as a consultant for Novo Nordisk. AS has served on advisory boards and/or lectured for MSD, AstraZeneca, Boehringer Ingelheim, Takeda, and Amgen AB. ML has received honoraria or been a consultant for AstraZeneca, Eli Lilly, and Novo Nordisk, participated in advisory boards for Novo Nordisk and received research grants from AstraZeneca and Dexcom outside the submitted work. The current study was an investigator‐initiated trial, where Novo Nordisk provided financial support and study drugs and Nordic InfuCare provided continuous glucose monitoring systems. Neither Novo Nordisk nor Nordic InfuCare had any role in the design or execution of the trial; the interpretation, analysis, or publication of data; or the decision to submit the written report.

Supporting information

 

Wijkman MO, Dena M, Dahlqvist S, et al. Predictors and correlates of systolic blood pressure reduction with liraglutide treatment in patients with type 2 diabetes. J Clin Hypertens. 2019;21:105–115. 10.1111/jch.13447