Effect of Hemoglobin A1c Reduction or Weight Reduction on Blood Pressure in Glucagon-Like Peptide-1 Receptor Agonist and Sodium-Glucose Cotransporter-2 Inhibitor Treatment in Type 2 Diabetes Mellitus: A Meta-Analysis.

Background Glucagon-like peptide-1 receptor agonists (GLP-1RAs) and sodium-glucose cotransporter-2 inhibitors (SGLT2is) have shown their beneficial effects on cardiovascular outcomes and multiple cardiovascular risk factors, including hypertension. However, the mechanism of blood pressure (BP)-lowering effects of these agents has not been elucidated. This study aims to evaluate the effect of hemoglobin A1c reduction or body weight reduction with GLP-1RA treatment and SGLT2i treatment on BP changes in patients with type 2 diabetes mellitus. Methods and Results Studies were identified by a search of MEDLINE, EMBASE, and the Cochrane Central Register until June 2019. Meta-regression analysis was performed to evaluate the association between hemoglobin A1c reduction or body weight reduction and changes of BP. A total of 184 trials were included. Both GLP-1RA and SGLT2i led to significant reductions in systolic BP (weighted mean difference, -2.856 and -4.331 mm Hg, respectively; P<0.001 for both) and diastolic BP (weighted mean difference, -0.898 and -2.279 mm Hg, respectively; P<0.001 for both). For both drug classes, hemoglobin A1c reduction was not independently associated with systolic BP reduction or diastolic BP reduction. In GLP-1RA treatment, weight reduction was positively associated with systolic BP reduction and diastolic BP reduction (β=0.821 and β=0.287, respectively; P<0.001 for both). In SGLT2i treatment, weight loss was significantly associated with systolic BP reduction (β=0.820; P=0.001) but was not associated with diastolic BP reduction. Conclusions Treatment with GLP-1RA and SGLT2i led to significant reductions in BP in patients with type 2 diabetes mellitus. Weight reduction was significantly and independently associated with BP reductions in GLP-1RA treatment and SGLT2i treatment.

diastolic blood pressure

glucagon‐like peptide‐1

glucagon‐like peptide‐1 receptor agonist

systolic blood pressure

sodium‐glucose cotransporter‐2 inhibitor

type 2 diabetes mellitus

weighted mean difference

Clinical Perspective

What Is New?

To date, 2 classes of antidiabetic agents, glucagon‐like peptide‐1 receptor agonist (GLP‐1RA) and sodium‐glucose cotransporter‐2 inhibitor (SGLT2i), have been shown to improve cardiovascular outcomes and multiple cardiovascular risk factors, including hypertension; however, the mechanism of blood pressure (BP)–lowering effects of these drugs has not been fully elucidated.

Evidence for the association between glycemic control or weight reduction and the BP changes provided a mix of results.

The effect of hemoglobin A1c reduction or weight reduction on BP changes in GLP‐1RA treatment and SGLT2i treatment is evaluated in this study.

What Are the Clinical Implications?

Weight reduction, not hemoglobin A1c reduction, is positively associated with BP reductions in GLP‐1RA treatment and SGLT2i treatment.

The findings of the present study might offer some insight into the potential mechanism by which GLP‐1RA and SGLT2i reduce BP in patients with diabetes mellitus.

Treatment with GLP‐1RA and SGLT2i results in weight loss and BP reduction in patients with diabetes mellitus, and these effects are attractive therapeutic properties in the management of type 2 diabetes mellitus.

Introduction

Type 2 diabetes mellitus (T2DM) is associated with a high risk of macrovascular events and microvascular disease.1, 2, 3 Hypertension is a common comorbidity that affects more than half of patients with T2DM and contributes to the risk of cardiovascular disease and microvascular complications.4, 5 It was demonstrated that optimal blood pressure (BP) control could reduce the risks of all‐cause mortality, cardiovascular disease, stroke, as well as diabetic retinopathy and albuminuria in patients with T2DM.6 BP control is therefore an important strategy for improving the prognosis of patients with T2DM.

Two classes of antidiabetic agents, glucagon‐like peptide‐1 receptor agonists (GLP‐1RAs)7, 8, 9, 10 and sodium‐glucose cotransporter‐2 inhibitors (SGLT2is),11, 12, 13 showed their beneficial effects on cardiovascular outcomes and multiple cardiovascular risk factors, including hypertension. The BP‐lowering effects of these 2 agents were established recently,14, 15, 16 but the exact mechanisms accounting for their antihypertensive effects were not elucidated yet. It was suggested that the BP reduction of GLP‐1RA treatment and SGLT2i treatment might be in part via weight loss.17, 18, 19 In addition, it was supposed that endothelial dysfunction and arterial stiffness induced by hyperglycemia might be involved in the pathogenesis of hypertension.20, 21 Thus, improvement in glycemic control may indirectly contribute to the BP‐lowering effect of these agents.

Previously, a pooled data analysis demonstrated that improved glycemic control and weight reduction was associated with BP reduction in patients with T2DM treated with exenatide.22 Furthermore, pooled analyses indicated that the weight loss associated with dapagliflozin and canagliflozin contributed to the reduction in systolic BP (SBP).23, 24 However, results from another study found a weak correlation between weight lost and reduction in SBP in exenatide‐treated patients.25 In addition, in a meta‐analysis evaluating the effects of SGLT2i on 24‐hour ambulatory BP, no significant association was observed between 24‐hour ambulatory BP and change in body weight.26 Some researchers indicated that the BP‐lowering effect occurred earlier than any significant weight loss in GLP‐1RA treatment27, 28, 29 and SGLT2i treatment,30 suggesting that the BP reduction may be mediated through mechanisms other than weight loss. To date, evidence for the association between blood glucose changes or weight reduction and the BP changes provided a mix of results.

Therefore, the aim of this meta‐analysis is to evaluate the effect of hemoglobin A1c (HbA1c) reduction or body weight reduction on BP changes in GLP‐1RA treatment and SGLT2i treatment in patients with T2DM.

Methods

The data that support the findings of this study are available from the corresponding author on reasonable request.

Search Strategy

This meta‐analysis was conducted according to the approach recommended by the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses guidelines.31 The study protocol is available in the International Prospective Register of Systematic Reviews, PROSPERO (registration Nos. CRD42018108738 and CRD42018105041).

Studies were identified by a literature search of MEDLINE, EMBASE, and the Cochrane Central Register of Controlled Trials until June 2019. The overall searching strategy was performed using T2DM separately with the following terms: GLP‐1RA, albiglutide, dulaglutide, exenatide, liraglutide, lixisenatide, semaglutide, taspoglutide, SGLT2i, dapagliflozin, canagliflozin, empagliflozin, ipragliflozin, tofogliflozin, luseogliflozin, ertugliflozin, remogliflozin, and sotagliflozin. Complete details of the search strategy are shown in Data S1.

Study Selection and Data Extraction

Studies were included if they met the following criteria: (1) randomized controlled trials comparing the efficacy and safety of GLP‐1RA or SGLT2i with placebo or other antidiabetic agents in participants with T2DM; (2) studies with duration ≥4 weeks; (3) the primary outcome was change in HbA1c, weight, or BP; cardiovascular outcome trials that reported BP changes from baseline were also included; and (4) studies that reported BP changes from baseline with mean difference.

The exclusion criteria were as follows: (1) non–randomized controlled trials; (2) studies in patients with type 1 diabetes mellitus; (3) studies with duration <4 weeks; (3) studies that did not report BP changes from baseline; and (4) studies that did not report SD, SE, or 95% CI of BP changes. Extension studies were excluded from this meta‐analysis to minimize the variations.

Two review authors (M.H. and S.Z.) independently performed the data extraction from each publication using a standardized form: publication data (study title, first author, publication year, and source of publication), study design, baseline characteristics of the study population (sample size, sex, age, body mass index [BMI], duration of T2DM, and baseline BP), description of the study drugs and dosages, duration of follow‐up, and changes of HbA1c, body weight, SBP, and diastolic BP (DBP) from baseline to study end point. Disagreements or discrepancies were resolved by discussion among the 2 review authors and a third investigator (W.Y.).

Assessment of Methodological Quality

The quality of each study was evaluated according to criteria provided in the Cochrane Handbook.32 Each trial was judged into low, high, or unclear risk of bias for the following aspects: sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessors, incomplete outcome data, selective outcome reporting, and other sources of bias.

Statistical Analysis

In this meta‐analysis, weighted mean difference (WMD) and 95% CI were calculated using inverse variance weighted random effect model to evaluate the changes of HbA1c, body weight, SBP, and DBP from baseline in GLP‐1RA and SGLT2i treatments. For placebo‐controlled trials and active‐controlled trials, variables compared with placebo or different classes of comparators were also calculated. Meta‐regression analysis was performed to evaluate the association between HbA1c reduction or body weight reduction and changes of BP. Confounding factors, including age, sex, BMI, and duration of diabetes mellitus, were adjusted by using multivariable meta‐regression analysis. Subgroup analyses were performed by pooling data for each individual GLP‐1RA and SGLT2i separately. If trials with >1 intervention group were identified, we determined which treatment groups in the study are relevant to our meta‐analysis/subgroup analysis, according to the Cochrane Handbook,32 and only these treatment groups were used in analyses. The number of observations refers to the number of treatment group (group of participants who receive GLP‐1RA or SGLT2i treatment) of studies. The heterogeneity among studies was assessed using the Higgins I2 statistics. Publication bias was assessed via a visual inspection of the funnel plot and Egger's test. All statistical analyses were conducted using STATA statistical software package, version 14.0.

Results

Search Selection and Characteristics

Details of the study selection process are presented by a flowchart (Figure 1). Finally, a total of 184 studies were included in the meta‐analysis, including 89 studies with GLP‐1RA treatment and 94 studies with SGLT2i treatment. One study compared the efficacy and safety of coinitiation of the GLP‐1RA and SGLT2i with either drug alone.33 A total of 44 trials compared a GLP‐1RA with a placebo, and 85 trials compared a SGLT2i with a placebo.

Figure 1

Flowchart of trial identification for meta‐analysis.

CENTRAL indicates Cochrane Central Register of Controlled Trials; GLP‐1RA, glucagon‐like peptide‐1 receptor agonist; and SGLT2i, sodium‐glucose cotransporter‐2 inhibitor.

This meta‐analysis was based on data from 61 299 individuals in the GLP‐1RA treatment and 40 874 individuals in the SGLT2i treatment. Characteristics of the individuals receiving GLP‐1RA treatment and SGLT2i treatment in this meta‐analysis were shown in Table S1. The range of age of the patients who received treatment with GLP‐1RA and SGLT2i was from 46.7 to 68.0 years and from 50.6 to 70.9 years, with the male percentage ranging from 24.8% to 83.0% and from 28.3% to 83.2%, respectively. The mean SBP level at randomization was 132.30 mm Hg (range, 122–138 mm Hg) in GLP‐1RA trials and 131.90 mm Hg (range, 122–151 mm Hg) in SGLT2i trials. Mean DBP level at baseline was 78.66 mm Hg (range, 72.9–84.8 mm Hg) in GLP‐1RA trials and 78.89 mm Hg (range, 73.5–91.2 mm Hg) in SGLT2i trials. Clinical characteristics of included studies with GLP‐1RA treatment and SGLT2i treatment are presented as Tables S2 and S3. The risk of bias was assessed with the Cochrane Handbook. Details of the quality of bias assessment were shown in Tables S4 and S5.

BP Changes in GLP‐1RA Treatment

Analysis of the pooled data across all studies showed that GLP‐1RA led to significant decreases in SBP (WMD, −2.856 mm Hg; 95% CI, −3.017 to −2.695 mm Hg; P<0.001) and DBP (WMD, −0.898 mm Hg; 95% CI, −1.007 to −0.789 mm Hg; P<0.001) from baseline (Figure 2 and Table S6). Compared with placebo, GLP‐1RA resulted in a significantly greater decrease in SBP (WMD, −1.724 mm Hg; 95% CI, −2.043 to −1.404 mm Hg; P<0.001). GLP‐1RA treatment was also associated with a significantly greater reduction in SBP in comparison with insulin treatment (WMD, −2.763 mm Hg; 95% CI, −3.306 to −2.220 mm Hg; P<0.001), sulfonylurea treatment (WMD, −2.721 mm Hg; 95% CI, −3.459 to −1.983 mm Hg; P<0.001), and dipeptidyl‐peptidase‐4 inhibitor treatment (WMD, −1.150 mm Hg; 95% CI, −1.657 to −0.644 mm Hg; P<0.001). No significant difference in DBP was found when GLP‐1RA treatment compared with placebo or active comparator treatment, except for sulfonylureas treatment (WMD, −1.318 mm Hg; 95% CI, −1.944 to −0.693 mm Hg; P<0.001). The changes in SBP and DBP with each individual GLP‐1RA treatment were also shown in Table S6. Subgroup analysis stratified by treatment strategy (monotherapy or combination therapy), study duration, and study primary end point showed that the results were similar as the total group, except for DBP changes in studies with BP change as the primary end point (WMD, 0.187 mm Hg; 95% CI, −0.471–0.845 mm Hg; P=0.577; Table S7), which may be attributed to the limited number of studies included in this subgroup. Statistical tests for the comparisons of the effect sizes among subgroups were shown in Table S8. HbA1c and weight changes in GLP‐1RA treatment were shown in Table S9.

Figure 2

Forest plots of systolic blood pressure (SBP ) changes (A) and diastolic blood pressure (DBP ) changes (B) in glucagon‐like peptide‐1 receptor agonist (GLP‐1RA ) treatment.

*Weighted mean difference (WMD) and 95% CI were calculated for a change from baseline to the study end point for GLP‐1RA vs placebo or different classes of comparators. DPP‐4i indicates dipeptidyl‐peptidase‐4 inhibitor; and HbA1c, hemoglobin A1c.

BP Changes in SGLT2i Treatment

Treatment with SGLT2i resulted in significant decreases in SBP (WMD, −4.331 mm Hg; 95% CI, −4.476 to −4.185 mm Hg; P<0.001) and DBP (WMD, −2.279 mm Hg; 95% CI, −2.376 to −2.182 mm Hg; P<0.001) from baseline (Figure 3 and Table S6). Compared with placebo, SGLT2i treatment led to a significantly greater reduction in SBP (WMD, −3.612 mm Hg; 95% CI, −3.844 to −3.379 mm Hg; P<0.001) and led to a significantly greater reduction in DBP (WMD, −1.559 mm Hg; 95% CI, −1.713 to −1.406 mm Hg; P<0.001). SGLT2i treatment was also associated with significantly greater decreases in SBP and DBP in comparison with metformin, sulfonylurea, and dipeptidyl‐peptidase‐4 inhibitor treatment. The changes in SBP and DBP with each individual SGLT2i treatment were also shown in Table S6. No significant differences in BP changes were found by subgroup analysis stratified by treatment strategy (monotherapy or combination therapy) and study duration. The effect of SGLT2i on SBP changes was greater in studies in which the primary end point was changes in BP (WMD, −6.331 mm Hg; 95% CI, −6.853 to −5.809 mm Hg; P<0.001; Table S7). The possible reason is that the baseline SBP levels of participants were higher in those studies. Statistical test for the comparisons of the effect sizes among subgroups were shown in Table S8. HbA1c and weight changes in SGLT2i treatment were shown in Table S10.

Figure 3

Forest plots of systolic blood pressure (SBP ) changes (A) and diastolic blood pressure (DBP ) changes (B) in sodium‐glucose cotransporter‐2 inhibitor (SGLT2i) treatment.

*Weighted mean difference (WMD ) and 95% CI were calculated for a change from baseline to the study end point for SGLT2i vs placebo or different classes of comparators. DPP‐4i indicates dipeptidyl‐peptidase‐4 inhibitor; and HbA1c, hemoglobin A1c.

Effect of HbA1c Change or Weight Reduction on BP Changes in GLP‐1RA Treatment

In terms of absolute BP changes, HbA1c change from baseline was significantly associated with SBP reduction (β=2.538; 95% CI, 1.652–3.425; P<0.001, adjusted for age, sex, BMI, and duration of diabetes mellitus; Figure 4A) and was also significantly associated with DBP reduction (adjusted β=0.727; 95% CI, 0.226–1.227; P=0.005; Figure 4B). In terms of placebo‐corrected BP changes, HbA1c reduction was positively associated with placebo‐corrected reduction in SBP (adjusted β=3.614; 95% CI, 2.107–5.122; P<0.001; Figure S1A) and HbA1c reduction was also positively associated with placebo‐corrected reduction in DBP (adjusted β=1.397; 95% CI, 0.280–2.515; P=0.015; Figure S1B).

Figure 4

Meta‐regression analysis of the associations between hemoglobin A1c (HbA1c) reduction or body weight reduction and blood pressure changes in glucagon‐like peptide‐1 receptor agonist (GLP‐1RA ) treatment.

A, Association between HbA1c change from baseline and systolic blood pressure (SBP ) change from baseline. B, Association between HbA1c change from baseline and diastolic blood pressure (DBP ) change from baseline. C, Association between weight change from baseline and SBP change from baseline. D, Association between weight change from baseline and DBP change from baseline. The size of circles is proportional to the weight of each study in the meta‐regression. *Analyses were adjusted for age, sex, body mass index (BMI), and duration of diabetes mellitus.

Weight change from baseline was significantly associated with net changes in SBP and DPB (adjusted β=0.904 [95% CI, 0.739–1.070] and adjusted β=0.296 [95% CI, 0.196–0.396], respectively; P<0.001 for both) in GLP‐1RA treatment (Figure 4C and 4D). In terms of placebo‐corrected BP changes, weight reduction was positively associated with placebo‐corrected SBP reduction with significance (adjusted β=0.523; 95% CI, 0.270–0.776; P<0.001), but was not associated with placebo‐corrected DBP reduction (Figure S1C and S1D). Details were shown in Table 1 and Figures S1 through S3.

Table 1

Effect of HbA1c Reduction or Weight Reduction on BP Changes in GLP‐1RA Treatment

Variable	SBP Changes			DBP Changes
	Coefficient	95% CI	P Value	Coefficient	95% CI	P Value
HbA1c change from baseline, %
Total	2.538	1.652, 3.425	<0.001	0.727	0.226, 1.227	0.005
Placebo controlled	3.614	2.107, 5.122	<0.001	1.397	0.280, 2.515	0.015
Active controlled	2.392	1.341, 3.443	<0.001	0.580	−0.005, 1.166	0.052
Insulin	1.255	−0.251, 2.761	0.096	0.327	−0.647, 1.302	0.484
Sulfonylurea	2.919	−9.527, 15.365	0.419	…	…	…
DPP‐4i	3.491	0.689, 6.292	0.018	0.228	−1.775, 2.232	0.810
Monotherapy	2.144	−0.414, 4.702	0.097	−0.092	−1.653, 1.469	0.904
Add‐on therapy	2.598	1.595, 3.600	<0.001	0.790	0.249, 1.331	0.005
Individual
Albiglutide	4.102	−6.201, 14.405	0.378	1.257	−3.472, 5.987	0.550
Dulaglutide	3.281	0.526, 6.037	0.022	0.570	−0.476, 1.616	0.271
Exenatide	2.105	−0.100, 4.309	0.061	0.362	−1.073, 1.797	0.608
Liraglutide	0.457	−1.453, 2.3681	0.626	0.065	−1.565, 1.695	0.934
Semaglutide	4.290	2.432, 6.148	<0.001	1.186	0.176, 2.196	0.023
Taspoglutide	−2.503	−6.628, 1.621	0.188	−0.141	−2.708, 2.426	0.898
Weight change from baseline, kg
Total	0.904	0.739, 1.070	<0.001	0.296	0.196, 0.396	<0.001
Placebo controlled	0.523	0.270, 0.776	<0.001	0.036	−0.130, 0.203	0.661
Active controlled	0.876	0.660, 1.093	<0.001	0.264	0.134, 0.395	<0.001
Insulin	0.403	−0.159, 0.965	0.147	0.109	−0.244, 0.463	0.518
Sulfonylurea	0.931	−3.359, 5.221	0.449	…	…	…
DPP‐4i	0.733	0.206, 1.259	0.010	0.087	−0.292, 0.466	0.631
Monotherapy	1.114	0.530, 1.698	0.001	0.342	−0.060, 0.744	0.092
Add‐on therapy	0.881	0.697, 1.064	<0.001	0.262	0.154, 0.371	<0.001
Individual
Albiglutide	1.481	−7.048, 10.010	0.694	1.621	−2.208, 5.450	0.350
Dulaglutide	0.710	−0.081, 1.501	0.076	0.170	−0.135, 0.474	0.260
Exenatide	1.811	1.155, 2.468	<0.001	0.610	0.053, 1.167	0.033
Liraglutide	0.277	−0.427, 0.981	0.425	0.151	−0.392, 0.695	0.565
Semaglutide	0.904	0.595, 1.214	<0.001	0.281	0.104, 0.457	0.003
Taspoglutide	0.654	−0.324, 1.632	0.153	0.160	−0.449, 0.769	0.544

Analyses were adjusted for age, sex, body mass index, and duration of diabetes mellitus by meta‐regression. BP indicates blood pressure; DBP, diastolic BP; DPP‐4i, dipeptidyl‐peptidase‐4 inhibitor; GLP‐1RA, glucagon‐like peptide‐1 receptor agonist; HbA1c, hemoglobin A1c; and SBP, systolic BP.

Effect of HbA1c Change or Weight Reduction on BP Changes in SGLT2i Treatment

HbA1c reduction was not associated with SBP reduction or DBP reduction in SGLT2i treatment (Figure 5A and 5B). In terms of absolute BP changes, weight reduction in SGLT2i was positively associated with SBP reduction with significance (adjusted β=0.771; 95% CI, 0.314–1.228; P=0.001), but was not associated with DBP reduction (Figure 5C and 5D). In addition, weight change from baseline was significantly associated with SBP reduction in SGLT2i monotherapy (adjusted β=1.211; 95% CI, 0.140–2.283; P=0.028), and weight reduction was also significantly associated with SBP reduction in SGLT2i add‐on therapy (adjusted β=0.711; 95% CI, 0.204–1.219; P=0.007). No significant association was observed between weight reduction and DBP reduction, as either monotherapy or add‐on therapy. In terms of placebo‐corrected BP changes, weight reduction was associated with placebo‐corrected reduction in SBP (adjusted β=0.965; 95% CI, 0.456–1.473; P<0.001; Figure S4C) and weight reduction was also associated with placebo‐corrected reduction in DBP (adjusted β=0.385; 95% CI, 0.042–0.728; P=0.028; Figure S4D). Details were shown in Table 2 and Figures S4 through S6.

Figure 5

Meta‐regression analysis of the associations between hemoglobin A1c (HbA1c) reduction or body weight reduction and blood pressure changes in sodium‐glucose cotransporter‐2 inhibitor (SGLT2i) treatment.

A, Association between HbA1c change from baseline and systolic blood pressure (SBP) change from baseline. B, Association between HbA1c change from baseline and diastolic blood pressure (DBP) change from baseline. C, Association between weight change from baseline and SBP change from baseline. D, Association between weight change from baseline and DBP change from baseline. The size of circles is proportional to the weight of each study in the meta‐regression. *Analyses were adjusted for age, sex, body mass index (BMI), and duration of diabetes mellitus.

Table 2

Effect of HbA1c Reduction or Weight Reduction on BP Changes in SGLT2i Treatment

Variable	SBP Changes			DBP Changes
	Coefficient	95% CI	P Value	Coefficient	95% CI	P Value
HbA1c change from baseline, %
Total	0.396	−0.745, 1.537	0.493	0.165	−0.517, 0.846	0.633
Placebo controlled	−0.237	−1.262, 0.788	0.647	−0.296	−1.005, 0.414	0.410
Active controlled	2.099	−0.256, 4.455	0.078	1.593	−0.036, 3.221	0.055
Metformin	…	…	…	…	…	…
Sulfonylurea	…	…	…	…	…	…
DPP‐4i	4.235	−0.735, 9.206	0.086	1.364	−3.183, 5.911	0.501
Monotherapy	0.587	−2.337, 3.510	0.684	0.597	−1.273, 2.467	0.516
Add‐on therapy	0.350	−0.850, 1.550	0.564	0.198	−0.546, 0.943	0.597
Individual
Canagliflozin	−0.412	−2.373, 1.549	0.669	−0.277	−1.275, 0.721	0.572
Dapagliflozin	−0.386	−2.179, 1.407	0.665	−0.303	−1.614, 1.008	0.638
Empagliflozin*	1.059	−0.889, 3.007	0.280	0.574	−0.667, 1.814	0.357
Ertugliflozin	0.870	−1.250, 2.990	0.394	1.891	−1.489, 5.272	0.233
Ipragliflozin	−3.190	−2.651, 2.173	0.773	−3.958	−45.880, 37.964	0.724
Luseogliflozin	13.433	−2.002, 28.868	0.073	6.465	−0.086, 13.016	0.052
Tofogliflozin	−6.672	−36.852, 23.509	0.442	−8.437	−28.661, 11.788	0.215
Weight change from baseline, kg
Total	0.771	0.314, 1.228	0.001	0.254	−0.016, 0.524	0.065
Placebo controlled	0.965	0.456, 1.473	<0.001	0.385	0.042, 0.728	0.028
Active controlled	0.924	0.052, 1.796	0.039	0.634	−0.035, 1.302	0.062
Metformin	…	…	…	…	…	…
Sulfonylurea	…	…	…	…	…	…
DPP‐4i	0.296	−1.504, 2.096	0.714	−0.519	−2.234, 1.196	0.487
Monotherapy	1.211	0.140, 2.283	0.028	0.767	−0.038, 1.572	0.061
Add‐on therapy	0.711	0.204, 1.219	0.007	0.238	−0.063, 0.540	0.119
Individual
Canagliflozin	0.763	−0.226, 1.752	0.125	0.343	−0.175, 0.861	0.184
Dapagliflozin	0.818	−0.054, 1.691	0.065	−0.143	−0.812, 0.526	0.662
Empagliflozin	0.292	−0.263, 0.847	0.295	0.327	−0.035, 0.689	0.076
Ertugliflozin	0.397	−0.642, 1.435	0.426	0.470	−0.847, 1.788	0.434
Ipragliflozin	1.681	−7.405, 10.766	0.597	0.421	−10.693, 11.536	0.885
Luseogliflozin	3.246	0.113, 6.378	0.045	1.365	−0.065, 2.795	0.057
Tofogliflozin	−0.534	−8.765, 7.698	0.806	−2.684	−8.459, 3.092	0.184

Analyses were adjusted for age, sex, body mass index, and duration of diabetes mellitus by meta‐regression. BP indicates blood pressure; DBP, diastolic BP; DPP‐4i, dipeptidyl‐peptidase‐4 inhibitor; HbA1c, hemoglobin A1c; SBP, systolic BP; and SGLT2i, sodium‐glucose cotransporter‐2 inhibitor.

For empagliflozin, analyses were adjusted for age, sex, and body mass index because most of the studies did not report duration of diabetes mellitus.

Effects of HbA1c and Weight Reduction on BP Changes in GLP‐1RA and SGLT2i Treatment

Analyses were conducted to explore the joint effects of HbA1c and weight reduction on BP changes (Table 3). In GLP‐1RA treatment, the associations between HbA1c reduction and SBP or DBP reduction became insignificant after further adjustment for weight change (Figure 4A and 4B). Weight reduction was positively associated with SBP reduction (β=0.821; 95% CI, 0.631–1.011; P<0.001), and weight reduction was also positively associated with DBP reduction (β=0.287; 95% CI, 0.172–0.403; P<0.001), independent of age, sex, BMI, duration of diabetes mellitus, and change in HbA1c (Figure 4C and 4D). In SGLT2i treatment, the effect of weight reduction on SBP change was also significant after adjustment for age, sex, BMI, duration of diabetes mellitus, and HbA1c change from baseline (β=0.820; 95% CI, 0.332–1.307; P=0.001; Figure 5C). Sex and antihypertensive therapy did not affect the association between weight loss and BP reductions in GLP‐1RA treatment and SGLT2i treatment (Tables S11 and S12). When data from GLP‐1RA studies and SGLT2i studies were merged into one data set, weight reduction was also positively and independently associated with SBP reduction and DBP reduction (β=0.903 [95% CI, 0.736–1.070] and β=0.375 [95% CI, 0.269–0.482], respectively; P<0.001 for both; Figure S7). Taken together, weight reduction was significantly and independently associated with BP reductions in GLP‐1RA treatment and SGLT2i treatment.

Table 3

Effects of HbA1c and Weight Reduction on BP Changes in GLP‐1RA and SGLT2i Treatment

Change From Baseline	SBP Changes			DBP Changes
	Coefficient	95% CI	P Value	Coefficient	95% CI	P Value
GLP‐1RA
HbA1c change, %	0.736	−0.111, 1.584	0.088	0.081	−0.454, 0.616	0.766
Weight change, kg	0.821	0.631, 1.011	<0.001	0.287	0.172, 0.403	<0.001
SGLT2i
HbA1c change, %	−0.337	−1.501, 0.826	0.567	−0.102	−0.820, 0.616	0.779
Weight change, kg	0.820	0.332, 1.307	0.001	0.268	−0.019, 0.556	0.067

Analysis was performed using meta‐regression, with age, sex, body mass index, duration of diabetes mellitus, HbA1c change from baseline, and weight change from baseline as covariates. BP indicates blood pressure; DBP, diastolic BP; GLP‐1RA, glucagon‐like peptide‐1 receptor agonist; HbA1c, hemoglobin A1c; SBP, systolic BP; and SGLT2i, sodium‐glucose cotransporter‐2 inhibitor.

Publication Bias

The funnel plots for SBP and DBP analysis in GLP‐1RA studies were symmetry (Figure S8), but Egger's regression analysis suggested the presence of publication bias in the analysis of DBP (Egger's test P=0.044). The funnel plot of SBP changes in SGLT2i studies showed slight asymmetry (Figure S9A), and Egger's regression analysis also detected a potential publication bias (P=0.025). No evidence of publication bias was found for DBP analysis in SGLT2i studies by funnel plot or Egger's test (P=0.682; Figure S9B). Imputation of possibly unpublished negative studies by trim‐and‐fill method34 did not significantly alter the general results, suggesting that the publication bias did not impact the interpretation of the results.

Discussion

To date, among various classes of antihyperglycemic agents, both GLP‐1RA and SGLT2i have been shown to improve cardiovascular outcomes in patients with T2DM.35, 36, 37 The cardiovascular benefits of these drugs may partly be attributable to their BP‐lowering effects. The present meta‐analysis showed that weight reduction, not HbA1c reduction, was significantly and independently associated with BP reductions in GLP‐1RA and SGLT2i treatment. These results indicated that weight loss contributed to the BP‐lowering effects of GLP‐1RA and SGLT2i.

The BP‐lowering effects of GLP‐1RA and SGLT2i have been well documented in clinical trials and previous meta‐analyses.38, 39, 40, 41, 42 It has been reported that GLP‐1RA treatment was associated with significant reductions in SBP and DBP in comparison with placebo or other antidiabetic drugs.16, 43 Similar results of favorable effects of SGLT2i on BP have also been reported in recent meta‐analyses and systematic reviews.26, 44, 45

The exact mechanism responsible for the BP‐lowering effects with these agents has not been fully understood. Weight loss may be one of the important factors because evidence in clinical trials and epidemiologic studies showed that weight loss was associated with reduced BP.46, 47, 48 A pooled analysis of 6 randomized controlled trials reported a weak correlation between weight loss and reduction in SBP for exenatide‐treated subjects.25 Similarly, a weak but statistically significant association between weight reduction and SBP lowering was observed in another pooled analysis of randomized controlled trials.49 Both studies showed that the SBP reduction in GLP‐1RA treatment weakly associated with weight loss, but both studies included only 6 trials and the relationship was calculated by linear correlation without adjusting for possible confounding factors. In addition, in a meta‐analysis of 33 GLP‐1RA trials, the degree of SBP change was not related to weight loss or improvement in HbA1c, but trials of patients without T2DM were also included.50 Paul et al have reported that short‐term dynamics of BP in exenatide treatment were related to concomitant effects on glycemia and body weight, demonstrating that improved glycemic control and weight reduction were associated with BP reduction in treatment with exenatide.22 Meta‐regression analysis in our study found that weight reduction was significantly associated with BP lowering in GLP‐1RA treatment, even after adjusting for possible confounding factors, including age, sex, BMI, duration of diabetes mellitus, and change in HbA1c, indicating that weight loss may contribute to the BP‐lowering effect of GLP‐1RA. However, results of the joint effects of HbA1c reduction and weight reduction on BP showed that HbA1c reduction was not correlated with BP changes after adjusting for weight loss. It is likely that glycemic control can be improved by weight loss; therefore, the effect of HbA1c reduction on BP changes may be dependent on reduction in weight.

Some researchers indicated that the BP reductions observed in the clinical trials occurred earlier than any significant weight loss, suggesting that GLP‐1RA treatment may provide extra benefits independent of weight loss that lead to BP lowering.51, 52 Vasodilatation and natriuresis mediated by activation of glucagon‐like peptide‐1 (GLP‐1) receptor on cardiovascular and renal tissue likely contribute to the antihypertensive effect.19, 53 A study found that infusion of recombinant GLP‐1 improved endothelial function in patients with T2DM and established coronary artery disease.54 Moreover, infusion of GLP‐1 enhanced acetylcholine‐mediated vasodilation.55 In patients with T2DM, the administration of exenatide was associated with increased plasma concentrations of a series of vasodilator and suppression of renin‐angiotensin system.56 These results indicated a potentially direct benefit on vascular factors of GLP‐1 in humans. On the other hand, sustained liraglutide administration increased urinary sodium excretion in hypertensive subjects with T2DM.57 Similarly, another study observed intravenous infusions of GLP‐1 promoted natriuresis in both healthy and insulin‐resistant obese men.58 Therefore, GLP‐1–induced natriuresis may provide another mechanism for antihypertensive effect associated with GLP‐1RA.

Several studies reported the association between weight reduction and BP changes in treatment with SGLT2i. A previous meta‐analysis that involved 6 trials reported SGLT2i significantly reduced 24‐hour ambulatory SBP and DBP. However, no significant association between change in body weight and 24‐hour BP was observed in the study.26 Pooled data from placebo‐controlled studies in patients with T2DM indicated that weight loss contributed to reductions in BP in treatment with dapagliflozin24 or canagliflozin.23 Data from our meta‐analysis also support the evidence that weight reduction was positively associated with BP reduction, independent of age, sex, BMI, duration of diabetes mellitus, and HbA1c reduction. However, reductions in BP in SGLT2i treatment were also observed before body weight reductions,30, 59 suggesting that the BP‐lowering effect of SGLT2i cannot solely be ascribed to weight loss. Osmotic diuresis and mild natriuresis are thought to be the most likely explanations for the antihypertensive effect of SGLT2i.59, 60, 61, 62 The glucose‐based osmotic diuresis leads to an excess urinary output by 110 to 470 mL/d.63 A 7% reduction in plasma volume was observed in patients with T2DM treated with dapagliflozin, indicating the diuretic‐like capacity of dapagliflozin possibly resulted from enhanced sodium excretion or osmotic diuresis.64 In addition, reduction in arterial stiffness induced by SGLT2i might also play a part in BP lowering.65, 66 Further studies are needed to elucidate the underlying mechanism by which GLP‐1RA and SGLT2i reduce BP in patients with T2DM.

In the current analysis, there was a greater effect of GLP‐1RA and SGLT2i on SBP compared with DBP. The differential effects may be attributed to the mechanism for the antihypertensive actions of both drugs. In the present study, we demonstrated that weight loss was associated with BP reductions in GLP‐1RA treatment and SGLT2i treatment. A difference in response in SBP compared with DBP to weight reduction was observed in previous meta‐analysis, in which the effects of weight loss appear to be larger on SBP than on DBP.46, 67 Moreover, the BP‐lowering effects of GLP‐1RA treatment and SGLT2i treatment are thought to be partly mediated through enhanced urinary sodium excretion. The magnitude of the association between serum sodium levels and SBP was greater than DBP.68 Wannamethee et al69 found a positive association between serum sodium and SBP in hypertensive individuals. Although there was also a slight tendency for DBP to increase with increasing serum sodium, the trend was not significant. Another possible explanation of these finding is that the plasma volume reduction resulting from the increase in urinary glucose excretion induced by SGLT2i62, 70 and the relative reduction in intravascular volume resulting from vasodilation induced by GLP‐1RA71, 72 would be more likely to result in reductions in SBP compared with DBP.

Our meta‐analysis involved a substantial number of placebo‐controlled trials and active‐controlled trials for GLP‐1RA or SGLT2i treatment. With data from 61 299 individuals in the GLP‐1RA treatment and 40 874 individuals in the SGLT2i treatment, our analysis provided sufficient power to evaluate the effect of HbA1c reduction or weight reduction on BP changes in patients with T2DM receiving GLP‐1RA treatment and SGLT2i treatment. However, we acknowledge several limitations of our study. First, there was some moderate level of heterogeneity across studies, which may influence the interpretation of the results. Data from separate studies were combined for analysis. Baseline characteristic, agent dosage, and duration of follow‐up varied across studies, which may cause a high level of heterogeneity. Confounding factors, such as the presence or absence of hypertension diagnosis in the study population, the background antihypertensive therapies, and changes in medication during the course of trial, were not available in many of the included studies, which might be another possible explanation for the heterogeneity. Second, most of the included studies were clinical assessment of efficacy of GLP‐1RA and SGLT2i treatment. Therefore, glycemic control was the primary end point in most of the studies and changes of BP were typically reported as safety outcomes or secondary outcomes. Third, the association examined by meta‐regression analysis may not be interpreted as a causal effect. Last, funnel plot analysis suggested the presence of publication bias. Although we used trim‐and‐fill method to further assess the impact of publication bias, our results should be interpreted with caution.

Conclusions

Treatment with GLP‐1RA and SGLT2i led to significant reductions in BP in patients with T2DM. Weight reduction was significantly and independently associated with BP reductions in GLP‐1RA treatment and SGLT2i treatment. These results indicated that weight loss contributed to the BP‐lowering effects of GLP‐1RA and SGLT2i. Further studies are needed to elucidate the underlying mechanism of the BP‐lowering effects of these 2 drugs and its potential impact on cardiovascular outcomes.

Sources of Funding

This meta‐analysis was supported by grants from the National Key Research and Development Program of China (2016YFC1304901) and the National Natural Science Foundation of China (81970698, 81970708). The funding agencies had no role in the study design, data collection or analysis, decision to publish, or preparation of the manuscript.

Disclosures

Dr Ji has received fees for lecture presentations from AstraZeneca, Merck, Novartis, Lilly, Roche, Sanofi‐Aventis, and Takeda. Dr Ji has received consulting fees from companies including AstraZeneca, Merck, Novartis, Lilly, Roche, Sanofi‐Aventis, and Takeda. Dr Ji has received grants/research support from AstraZeneca, Bristol‐Myers Squibb, Merck, Novartis, and Sanofi‐Aventis. The remaining authors have no disclosures to report.

Data S1

Tables S1–S12

Figures S1–S9

References 9, 11, 13, and 73–240

Click here for additional data file.