Characteristics of participants
This study included 615 T2DM participants with 411 (66.8%) male and a mean age of 49.2 ± 12.1 years. The mean HbA1c for all participants was 9.8 ± 2.1% (83.6 ± 23.4 mmol/mol) and the median duration of diabetes was 3.7 (0.1, 10.0) years. Compared to those with HbA1c < 10%, participants with HbA1c ≥10% were younger (46.3 ± 12.3 years vs. 51.5 ± 11.4 years, P < 0.01), had a shorter duration of diabetes [1.0 (0.1, 7.0) years vs. 5.6 (1.3, 10.7) years, P < 0.01], a lower prevalence of self-reported hypertension (27.6% vs. 35.6%, P < 0.05), coronary heart disease (8.5% vs. 20.3%, P < 0.01), stroke (6.4% vs. 12.8%, P < 0.01), and a lower level of beta-cell function [97.1 (66.5, 136.5) vs. 141.0 (87.9, 198.0), P < 0.01] (Table 1).
Table 1
Demographic and clinical characteristics of participants
Characteristics
|
All participants
|
Participants with HbA1c < 10%
|
Participants with HbA1c ≥10%
|
P value*
|
N
|
615
|
320
|
283
|
|
Age, mean ± SD, year
|
49.2 ± 12.1
|
51.5 ± 11.4
|
46.3 ± 12.3
|
< 0.01
|
Male sex, n (%)
|
411 (66.8)
|
206 (64.4)
|
197 (69.6)
|
0.17
|
Current smoking, n (%)
|
198 (32.4)
|
95 (29.9)
|
102 (36.2)
|
0.10
|
Current alcohol consumption, n (%)
|
280 (45.8)
|
138 (43.4)
|
139 (49.3)
|
0.15
|
Duration of diabetes, median (25th, 75th), year
|
3.7 (0.1, 10.0)
|
5.6 (1.3, 10.7)
|
1.0 (0.1, 7.0)
|
< 0.01
|
Self-reported hypertension, n (%)
|
198 (32.2)
|
114 (35.6)
|
78 (27.6)
|
0.03
|
CHD, n (%)
|
92 (15.0)
|
65 (20.3)
|
24 (8.5)
|
< 0.01
|
Stroke, n (%)
|
62 (10.1)
|
41 (12.8)
|
18 (6.4)
|
< 0.01
|
BMI, mean ± SD, kg/m2
|
25.9 ± 3.7
|
26.1 ± 3.9
|
25.7 ± 3.6
|
0.31
|
SBP, mean ± SD, mmHg
|
131.3 ± 17.2
|
133.0 ± 17.2
|
129.2 ± 17.0
|
< 0.01
|
DBP, mean ± SD, mmHg
|
80.9 ± 10.5
|
81.3 ± 10.5
|
80.4 ± 10.6
|
0.29
|
HbA1c, mean ± SD, %
|
9.8 ± 2.1
|
8.2 ± 1.2
|
11.6 ± 1.3
|
< 0.01
|
HbA1c, mean ± SD, mmol/mol
|
83.6 ± 23.4
|
65.8 ± 12.7
|
103.7 ± 14.6
|
< 0.01
|
FBG, mean ± SD, mmol/L
|
8.3 ± 2.3
|
7.6 ± 2.0
|
8.9 ± 2.5
|
< 0.01
|
2h PBG, mean ± SD, mmol/L
|
17.0 ± 4.6
|
16.0 ± 4.5
|
18.0 ± 4.5
|
< 0.01
|
TG, median (25th, 75th), mmol/L
|
1.5 (1.0, 2.5)
|
1.5 (1.0, 2.4)
|
1.5 (1.0, 2.7)
|
0.68
|
TC, median (25th, 75th), mmol/L
|
4.5 (3.8, 5.2)
|
4.4 (3.8, 5.1)
|
4.7 (4.0, 5.4)
|
< 0.01
|
HDL-C, median (25th, 75th), mmol/L
|
1.0 (0.8, 1.2)
|
1.0 (0.8, 1.2)
|
1.0 (0.8, 1.2)
|
0.96
|
LDL-C, median (25th, 75th), mmol/L
|
2.6 (2.0, 3.2)
|
2.5 (2.0, 3.1)
|
2.8 (2.2, 3.4)
|
< 0.01
|
ALT, median (25th, 75th), U/L
|
19.1 (13.3, 29.4)
|
19.1 (13.3, 29.4)
|
18.9 (13.4, 28.3)
|
0.82
|
AST, median (25th, 75th), U/L
|
15.9 (12.7, 21.8)
|
16.3 (13.4, 22.0)
|
15.1 (12.0, 20.2)
|
< 0.01
|
UA, median (25th, 75th), µmol/L
|
286.0 (232.0, 348.0)
|
295.0 (245.0, 356.5)
|
274.0 (216.0, 334.0)
|
< 0.01
|
Oral antihypertensive agent treatment, n (%)
|
137 (22.3)
|
86 (26.9)
|
49 (17.3)
|
< 0.01
|
Insulin treatment, n (%)
|
140 (22.8)
|
80 (25.0)
|
54 (19.1)
|
0.08
|
Oral hypoglycemic agent treatment, n (%)
|
382 (62.1)
|
236 (73.8)
|
137 (48.4)
|
< 0.01
|
HOMA2-IR, median (25th, 75th)
|
1.7 (1.3, 2.3)
|
1.8 (1.3, 2.4)
|
1.7 (1.4, 2.2)
|
0.11
|
Matsuda index, median (25th, 75th)
|
81.3 (54.3, 120.4)
|
74.7 (48.4, 106.5)
|
90.2 (63.5, 133.7)
|
< 0.01
|
HOMA2-B, median (25th, 75th)
|
51.5 (36.3, 80.0)
|
64.7 (41.7, 95.7)
|
44.7 (30.8, 61.0)
|
< 0.01
|
ISSI2, median (25th, 75th)
|
114.8 (77.0, 172.7)
|
141.0 (87.9, 198.0)
|
97.1 (66.5, 136.5)
|
< 0.01
|
*Compared the difference between two HbA1c groups. P value was from two independent sample t-test, Wilcoxon rank sum test, or Chi-square test. CHD, coronary heart disease; BMI, body mass index; SBP, systolic blood pressure; DBP, diastolic blood pressure; HbA1c, glycosylated hemoglobin; FBG, fasting blood glucose; PBG, postprandial blood glucose; TG, triglyceride; TC, total cholesterol; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; ALT, alanine transaminase; AST, aspartate aminotransferase; UA, uric acid; HOMA2-IR, homeostatic model assessment 2-insulin resistance; HOMA2-B, homeostatic model assessment 2-beta; ISSI2, insulin secretion-sensitivity index-2.
|
Higher Sbp Was Associated With Better Beta-cell Function In Participants With Hba1c ≥10%
To examine whether a relatively higher BP level might enhance the beta-cell function of T2DM patients with high glycemic level, we performed the analysis in participants with HbA1c < 10% and ≥10%, respectively. The characteristics of participants with different SBP in HbA1c < 10% or ≥10% were demonstrated in Supplementary Table 1. The level of HbA1c, FBG, 2h PBG, TG, total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), and LDL-C were not significantly different among these SBP groups. Using the multivariable linear regression analysis, we demonstrated that, for participants with HbA1c ≥10%, SBP was positively associated with log ISSI2 (b = 0.044, 95%CI: 0.011, 0.078) after adjustment of age, sex, duration of diabetes, log TG, LDL-C, log Matsuda index, oral antihypertensive agent treatment, insulin treatment, and oral hypoglycemic agent treatment (Table 2).
To examine whether there was a favorable SBP range corresponding to the better islet beta-cell function, we divided the participants into different SBP groups and compared their ISSI2. We found in participants with HbA1c ≥10%, SBP between 140–150 mmHg had the highest log ISSI2 (b = 0.306, 95%CI: 0.126, 0.485). However, this phenomenon was not found in participants with HbA1c < 10% (Table 2). To evaluate the robustness of the results in participants with HbA1c ≥10%, we conducted sensitivity analyses in the following subjects: 1. insulin sensitivity in 10th-90th percentiles; 2. without taking oral antihypertensive agents; 3. without accepting insulin treatment; 4. duration of diabetes < 5 years. We found the results from above subjects all supported that SBP of 140–150 mmHg was associated with the better log ISSI2 (Supplementary Table 2).
Table 2
Association between SBP and ISSI2 in participants with HbA1c < 10% or ≥10%
Factor
|
Participants with HbA1c < 10%
|
Participants with HbA1c ≥10%
|
N
|
ISSI2
|
b (95%CI)
|
N
|
ISSI2
|
b (95%CI)
|
Continuous variable
|
|
|
|
|
|
|
SBP, per 10 mmHg
|
308
|
141.3 (89.2, 198.7)
|
-0.009 (-0.043, 0.025)
|
271
|
97.6 (66.5, 138.9)
|
0.044 (0.011, 0.078)
|
Categorical variable
|
|
|
|
|
|
|
90≤SBP < 120, mmHg
|
73
|
150.4 (95.3, 206.9)
|
Ref.
|
87
|
90.7 (67.6, 127.9)
|
Ref.
|
120≤SBP < 130, mmHg
|
62
|
143.5 (85.5, 196.2)
|
0.010 (-0.162, 0.182)
|
64
|
110.6 (75.1, 150.9)
|
0.163 (0.015, 0.311)
|
130≤SBP < 140, mmHg
|
71
|
145.5 (108.0, 213.1)
|
0.064 (-0.102, 0.231)
|
54
|
95.9 (63.6, 135.2)
|
0.135 (-0.022, 0.292)
|
140≤SBP < 150, mmHg
|
50
|
140.5 (86.3, 185.8)
|
-0.076 (-0.259, 0.107)
|
35
|
108.6 (65.3, 188.2)
|
0.306 (0.126, 0.485)
|
150≤SBP, mmHg
|
52
|
109.3 (86.2, 179.1)
|
-0.016 (-0.201, 0.169)
|
31
|
91.8 (60.2, 132.8)
|
0.227 (0.028, 0.427)
|
Data of ISSI2 were median (25th, 75th). b(95%CI) was from linear regression analysis. ISSI2 was analyzed in models with the form of log transformation. All models adjusted age, sex, duration of diabetes, log TG, LDL-C, log Matsuda index, oral antihypertensive agent treatment, insulin treatment, and oral hypoglycemic agent treatment. ISSI2, insulin secretion-sensitivity index-2; SBP, systolic blood pressure; HbA1c, glycosylated hemoglobin; TG, triglyceride; LDL-C, low-density lipoprotein cholesterol.
|
To further identify the optimal SBP level corresponding to the best beta-cell function, we performed RCS analyses in participants with HbA1c < 10% and ≥10%, respectively. In participants with HbA1c ≥10%, we observed an inverted U-shape association between SBP and beta-cell function. Notably, ISSI2 increased with SBP increasing until SBP reached 146 mmHg (difference between 146 mmHg and Ref.= 30.93, 95%CI: 13.35, 48.51), then ISSI2 tended to decrease with SBP increasing (Fig. 1A). However, neither a linear nor a non-linear association was observed in participants with HbA1c < 10% (Fig. 1B).
Higher Sbp Was “paradoxically” Associated With Lower 2h Pbg In Participants With Hba1c ≥10%
The above results indicated that a certain degree of elevated SBP was associated with well-performed beta-cell function in participants with HbA1c ≥10%, then we try to determine whether blood glucose could be improved with SBP increasing. We performed the RCS analysis on the association of SBP with 2h PBG or FBG. We found SBP was non-linearly associated with 2h PBG in participants with HbA1c ≥10%. More precisely, 2h PBG decreased as SBP increased from 120 mmHg to 147 mmHg (difference between 147 mmHg and Ref.=-2.73 mmol/L, 95%CI: -3.97, -1.50), then 2h PBG tended to rise as SBP continued to increase (Fig. 2A). Whereas, neither a linear nor a non-linear association was observed between SBP and 2h PBG in participants with HbA1c < 10% (Fig. 2B). Similarly, SBP was not linearly or non-linearly associated with FBG either in participants with HbA1c ≥10% (Fig. 2C) or < 10% (Fig. 2D).
Beta-cell Function Mediated The Association Between Sbp And 2h Pbg In Participants With Hba1c ≥10%
The above results demonstrated that SBP around less than 150 mmHg was associated with increased ISSI2 and decreased 2h PBG in participants with HbA1c ≥10%. Then, to examine whether the association between SBP and 2h PBG was mediated by the beta-cell function, we performed mediation analyses in participants with the SBP < 150 mmHg. Mediation analyses showed that SBP was negatively associated with 2h PBG before adjusting for ISSI2 (total effect=-0.062, 95%CI: -0.102, -0.022) but not after adjusting for ISSI2 (direct effect=-0.005, 95%CI: -0.035, 0.025). The mediating effect of ISSI2 was − 0.057 (95%CI: -0.085, -0.029), which indicated that beta-cell function may mediate 92% (P < 0.01) of the association between SBP and 2h PBG (Fig. 3A). Furthermore, to avoid reverse causality, we repeated the analysis with SBP as the dependent variable, 2h PBG as the independent variable, and ISSI2 as the potential mediating variable. The direct effect of 2h PBG on SBP was 0.103 (95%CI: -0.448, 0.654) and the mediating effect of ISSI2 was − 0.363 (95%CI: -0.728, 0.002), which indicated that 2h PBG may not influence SBP and beta-cell function may not be a mediator in this analysis (Fig. 3B).