Cohort study
A total of 18,601 patients with T2DM undergoing PCI/CAG/CCTA were admitted from December 2017 to July 2024 in Beijing Hospital. Out of the admitted patients, a total of 2,350 were taking dapagliflozin and and 16,251 did not take any SGLT2i. After PSM, 2.071 cases and 2,071 controls were finally paired. The selection process is summarized in Fig. 1.
Baseline demographic and clinical characteristics before and after matching are described in Table 1 and Table 2. Each patient taking dapagliflozin was matched to a patient without taking any SGLT2i using clinical baseline variables and/or factors that may affect renal function. After PSM, the 2,071 matched pair of patients showed no significant difference in baseline characteristics such as age, gender, and BMI. There was no statistically significant difference in eGFR, BUN, AST, ALT, and others between the two groups either.
Table 1
Basic characteristics of patients in two groups before propensity matching.
Parameter | DAPA users (n = 2,350) | Control (n = 16,251) | P-value |
Baseline characteristics |
Age, years (IQR) | 66 (59–72) | 66 (59–73) | 0.077 |
Female gender, n (%) | 601 (25.57) | 5,110 (31.44) | 0.078 |
BMI, kg/m2 | 25.69 (23.70−27.99) | 25.54 (23.44–27.78) | 0.230 |
SBP, mmHg | 135 (120–148) | 135 (121–148) | 0.979 |
DBP, mmHg | 79 (68–86) | 79 (68–87) | 0.255 |
HR, bpm | 74 (66–82) | 74 (65–83) | 0.281 |
Comorbidities |
Smoking, n (%) | 931 (39.62) | 7,056 (43.42) | 0.049 |
Drinking, n (%) | 790 (33.62) | 6,073 (37.37) | 0.211 |
Hypertension, n (%) | 1,274 (54.21) | 10,194 (62.74) | 0.884 |
Dyslipidemia, n (%) | 1,481 (63.02) | 12,088 (74.38) | 0.119 |
COPD, n (%) | 131 (5.57) | 909 (5.59) | 0.123 |
Angina, n (%) | 664 (28.26) | 4,794 (29.50) | 0.004 |
Prior PCI, n (%) | 402 (17.11) | 2,851 (17.54) | 0.015 |
Prior CABG, n (%) | 29 (1.23) | 228 (1.40) | 0.939 |
Prior MI, n (%) | 288 (12.26) | 2,026 (12.47) | 0.032 |
Prior CI, n (%) | 70 (2.98) | 492 (3.03) | 0.312 |
AF, n (%) | 63 (2.68) | 377 (2.32) | 0.032 |
HF, n (%) | 116 (4.94) | 749 (4.61) | 0.029 |
Laboratory variables |
eGFR, ml/min/1.73 m2 | 86.20 (67.73−106.31) | 85.41 (66.95−105.49) | 0.109 |
HbA1c, % | 6.80 (6.20–7.80) | 7.10 (6.10–8.10) | 0.097 |
hs-TNI, pg/mL | 4.50 (0.13–12.70) | 0.55 (0.01−7.00) | 0.003 |
CK, U/L | 75.00 (56.00−105.00) | 73.00 (52.00−107.00) | 0.061 |
CK-MB, U/L | 1.30 (0.90–2.10) | 1.40 (0.90–2.30) | 0.136 |
Total cholesterol, mg/dL | 134.38 (111.76−161.64) | 134.96 (113.69−162.03) | 0.069 |
LDL-c, mg/dL | 74.63 (57.62–97.84) | 75.02 (56.84−99.00) | 0.085 |
HDL-c, mg/dL | 40.22 (34.42–47.18) | 40.22 (34.42–47.56) | 0.328 |
Triglyceride, mg/dL | 45.24 (31.71–63.03) | 47.18 (34.03–66.51) | 0.182 |
Hemoglobin, g/L | 16.00 (124.00−146.00) | 135.08 ± 16.58 | 0.127 |
C-reactive protein, mg/L | 0.68 (0.30–2.10) | 0.60 (0.20–1.90) | 0.121 |
BUN, mg/dL | 15.49 (12.60−19.16) | 15.23 (12.54–18.68) | 0.078 |
ALT, U/L | 18.00 (13.00–26.00) | 18.00 (13.00–26.00) | 0.545 |
AST, U/L | 19.00 (16.00–24.00) | 18.00 (15.00–23.00) | 0.207 |
Medications, n (%) |
Metformin | 862 (36.68) | 6,808 (41.89) | 0.704 |
Sulfonylureas | 129 (5.49) | 1,160 (7.14) | 0.183 |
DPP−4i | 278 (11.83) | 1,934 (11.90) | 0.022 |
GLP−1RA | 87 (3.70) | 512 (3.15) | 0.007 |
Glitazone | 56 (2.38) | 339 (2.09) | 0.052 |
Insulin | 460 (19.57) | 3206 (19.73) | 0.002 |
Anti platelets | 1,876 (79.83) | 14,734 (90.67) | 0.044 |
Anti coagulation | 47 (2.00) | 259 (1.59) | 0.018 |
ACEI | 98 (4.71) | 953 (5.86) | 0.055 |
ARB | 1,096 (46.64) | 7,113 (43.77) | < 0.000 |
β-blockers | 1,165 (49.57) | 8,129 (50.02) | |
CCB | 701 (29.83) | 5,641 (34.71) | 0.802 |
Diuretics | 463 (19.70) | 3,021 (18.59) | < 0.000 |
Statins | 1,845 (78.51) | 14,280 (87.87) | 0.001 |
Ezetimibe | 500 (21.28) | 3,082 (18.96) | < 0.000 |
DAPA, Dapagliflozin; BMI, Body Mess Index; SBP, Systolic Blood Pressure; DBP, Diastolic Blood Pressure; HR, Heart Rate; COPD, Chronic Obstructive Pulmonary Disease; PCI, Percutaneous Coronary Intervention; CABG, Coronary Artery Bypass Grafting; MI, Myocardial Infarction; CI, Cerebral Infarction; AF, Atrial Fibrillation; HF, Heart Failure; eGFR, estimated Glomerular Filtration Rate; HbA1c, glycated Hemoglobin A1c; hs-TnI, high-sensitivity Troponin I; CK, Creatine Kinase; CK-MB, Creatine Kinase-MB; LDL-c, Low-Density Lipoprotein Cholesterol; HDL-c, High-Density Lipoprotein Cholesterol; BUN, Blood Urea Nitrogen; ALT, Alanine Aminotransferase; AST, Aspartate Aminotransferase; DPP-4i, Dipeptidyl Peptidase-4 inhibitor; GLP-1RA, Glucagon-like Peptide-1 Receptor Agonist; ACEI, Angiotensin-Converting Enzyme Inhibitor; ARB, Angiotensin Receptor Blocker; CCB, Calcium Channel Blocker. |
Table 2
Basic characteristics of patients in two groups in propensity-matched dataset.
Parameter | DAPA users (n = 2,071) | Control (n = 2,071) | P-value |
Baseline characteristics |
Age, years | 66 (59–72) | 66 (58–72) | 0.490 |
Female gender, n (%) | 601 (29.02) | 549 (26.51) | 0.071 |
BMI, kg/m2 | 26.69 (23.74–27.99) | 25.71 (23.59–27.99) | 0.620 |
SBP, mmHg | 135 (121−148.5) | 134 (120–148) | 0.999 |
DBP, mmHg | 79 (69–86) | 79 (67–87) | 0.108 |
HR, bpm | 74 (66–82) | 74 (64–83) | 0.863 |
Comorbidities |
Smoking, n (%) | 931 (44.95) | 964 (46.55) | 0.303 |
Drinking, n (%) | 790 (38.15) | 832 (40.17) | 0.181 |
Hypertension, n (%) | 1,274 (61.52) | 1,232 (59.49) | 0.182 |
Dyslipidemia, n (%) | 1,481 (71.51) | 1,521 (73.44) | 0.164 |
COPD, n (%) | 131 (6.33) | 108 (5.21) | 0.125 |
Angina, n (%) | 664 (32.06) | 650 (31.39) | 0.640 |
Prior PCI, n (%) | 402 (19.41) | 374 (18.06) | 0.265 |
Prior CABG, n (%) | 29 (1.40) | 28 (1.35) | 0.894 |
Prior MI, n (%) | 288 (13.91) | 238 (11.49) | 0.020 |
Prior CI, n (%) | 70 (3.38) | 60 (2.90) | 0.373 |
AF, n (%) | 63 (3.04) | 56 (2.70) | 0.515 |
HF, n (%) | 116 (5.60) | 92 (4.44) | 0.088 |
Laboratory variables |
eGFR, ml/min/1.73 m2 | 86.17 (67.28−106.43) | 87.19 (68.53−107.65) | 0.647 |
HbA1c, % | 6.90 (6.20–7.90) | 7.10 (6.20–8.10) | 0.066 |
hs-TNI, pg/mL | 4.80 (1.04–13.59) | 0.94 (0.01–7.59) | 0.063 |
CK, U/L | 75.00 (56.00−104.00) | 73.00 (53.00−109.00) | 0.045 |
CK-MB, U/L | 1.30 (0.90–2.10) | 1.40 (0.90–2.30) | 0.229 |
Total cholesterol, mg/dL | 133.02 (110.60−160.09) | 134.18 (112.92−161.64) | 0.700 |
LDL-c, mg/dL | 73.86 (56.46–96.49) | 75.02 (56.84−99.00) | 0.595 |
HDL-c, mg/dL | 39.83 (34.42–46.79) | 39.44 (33.64–47.18) | 0.233 |
Triglyceride, mg/dL | 45.63 (32.10−63.23) | 47.56 (34.80–66.90) | 0.049 |
Hemoglobin, g/L | 137.00 (124.50–147.00) | 137.00 (126.00−148.00) | 0.547 |
C-reactive protein, mg/L | 0.70 (0.30–2.19) | 0.60 (0.20−2.00) | 0.408 |
BUN, mg/dL | 15.74 (12.85–19.54) | 15.34 (12.67–18.83) | 0.076 |
ALT, U/L | 18.00 (13.00–26.00) | 18.00 (13.00–26.00) | 0.775 |
AST, U/L | 19.00 (16.00–24.00) | 18.00 (15.00–23.00) | 0.294 |
Medications, n (%) |
Metformin | 862 (41.62) | 856 (41.33) | 0.850 |
Sulfonylureas | 129 (6.23) | 138 (6.66) | 0.569 |
DPP−4i | 278 (13.42) | 222 (10.72) | 0.008 |
GLP−1RA | 87 (4.20) | 68 (3.28) | 0.120 |
Glitazone | 56 (2.70) | 35 (1.69) | 0.026 |
Insulin | 460 (22.21) | 419 (20.23) | 0.119 |
Anti platelets | 1,876 (90.58) | 1,852 (89.43) | 0.214 |
Anti coagulation | 47 (2.27) | 42 (2.03) | 0.592 |
ACEI | 98 (4.73) | 110 (5.31) | 0.393 |
ARB | 1096 (52.92) | 905 (43.70) | < 0.000 |
β-blockers | 1,165 (56.25) | 1,019 (49.20) | < 0.000 |
CCB | 701 (33.85) | 694 (33.51) | 0.818 |
Diuretics | 463 (22.36) | 372 (17.96) | < 0.000 |
Statins | 1845 (89.09) | 1,793 (86.58) | 0.013 |
Ezetimibe | 500 (24.14) | 386 (18.64) | < 0.000 |
Before PSM, the unadjusted ORs of patients with CI-AKIESUR were 59.8% lower in the dapagliflozin user group [OR 0.402, 95%CI 0.322–0.501, P < 0.000] compared with the control group. Correlation analysis using KDIGO definition also rendered similar results [OR 0.306, 95% CI 0.235–0.399, P < 0.000]. After PSM, the adjusted OR of CI-AKIESUR remained 57.3% lower in the dapagliflozin group [OR 0.427, 95% CI 0.329–0.554, P < 0.000] than the control group. We also conducted subgroup analyses for different operation methods, including PCI, CAG and CCTA. The unadjusted and adjusted point estimates were qualitatively similar to the overall results. Results are shown in Table 3.
Table 3
Renal outcomes in the dapagliflozin group and control group.
Parameter | Unmatched population | Matched population |
OR | 95% CI | P-value | OR | 95% CI | P-value |
CI-AKIESUR | | | | | | |
Total | 0.402 | 0.322–0.501 | < 0.000 | 0.427 | 0.329–0.554 | < 0.000 |
PCI | 0.432 | 0.320–0.582 | < 0.000 | 0.503 | 0.348–0.726 | < 0.000 |
CAG | 0.384 | 0.275–0.535 | < 0.000 | 0.381 | 0.261–0.559 | < 0.000 |
CCTA | 0.125 | 0.016–0.947 | < 0.000 | 0.119 | 0.013–1.067 | 0.029 |
CI-AKIKDIGO | | | | | | |
Total | 0.306 | 0.235–0.399 | < 0.000 | 0.316 | 0.234–0.427 | < 0.000 |
PCI | 0.319 | 0.222–0.458 | < 0.000 | 0.354 | 0.232–0.540 | 0.006 |
CAG | 0.310 | 0.210–0.458 | < 0.000 | 0.306 | 0.197–0.473 | < 0.000 |
CCTA | 0.833 | 0.780–0.891 | < 0.000 | 0.408 | 0.291–0.572 | < 0.000 |
CAG, Coronary Angiography; CCTA, Coronary Computed Tomography Angiography; OR, Odd Ratio; CI, Confidence Interval. |
Meta-analysis of cohorts
Study Selection. Including the present study, we considered 6 studies with up to a total of 1,933 T2DM patients for meta-analysis. Data extraction flow was detailed in Additional File: PRISMA 2020 flow diagram. Initially, we conducted a meta-analysis, including 6 cohort studies.13, 14, 19–22
Study Characteristics. The characteristics of the included studies are provided in Additional File: Supplementary File eTable 2. Of the 6 studies in this meta-analysis, two were prospective cohort studies (PCS)19, 13 and four were retrospective cohort studies (RCS).13, 14, 19, 22 Regarding outcome treatment, one study used both ESUR and KDIGO criteria simultaneously.13 Two studies used the ESUR criteria.14, 19 One article used the KDIGO criteria.21 Another two articles adopted other standards.20, 22
GRADE assessment. We upgraded the level of CoE as all the studies included in the meta-analysis showed a low risk of bias. For details, please refer to the Additional File: Supplementary File eTable 3. Indirectness (the included studies compared similar interventions, similar populations, and similar outcomes), imprecision (this meta-analysis included 6,075 patients with diabetes undergoing CAG or PCI, 2,735 SGLT2i users, and 3,340 events of CI-AKI), publication bias, and inconsistency (I2 = 0) did not impact significantly the CoE. We assessed the CoE according to GRADE criteria as moderate.
Meta-analysis. Due to the different definition criteria of the outcome indicator CI-AKI in the included studies, some studies selected both ESUR and KDIGO criteria. This meta-analysis conducted two sets of analyses. In one set of analyses, when a study had both standards simultaneously, the results corresponding to the ESUR standard were selected for inclusion in the meta-analysis (Primary Outcome A). In the other set of analyses, when a study had both standards simultaneously, the results corresponding to the KDIGO criteria were selected for inclusion in the meta-analysis (Primary Outcome B).
Primary Outcomes. Among 6,075 patients in the 7 cohort studies, the use of SGLT2i were associated with significantly reduced CI-AKI outcomes [Primary Outcome A: RR 0.42, 95%CI 0.35–0.52, P < 0.000; Primary Outcome B: RR 0.37, 95%CI 0.30–0.45, P < 0.000] in T2DM patients after PCI/CAG/CCTA. The results are shown in detail in Fig. 2A and B.
Secondary Outcomes. In this meta-analysis, four included studies used the ESUR criteria for result statistics. A subgroup analysis was conducted based on the outcome treatment with the ESUR criteria. The results showed that SGLT2i could significantly reduce the incidence of CI-AKIESUR [RR 0.43, 95%CI 0.35–0.53, P < 0.000]. Three included studies used the KDIGO criteria for result statistics. A subgroup analysis was conducted based on the outcome treatment with the KDIGO criteria, and the results indicated that SGLT2i could significantly reduce the incidence of CI-AKIKDIGO [RR 0.36, 95%CI 0.28–0.47, P < 0.000]. Please refer to Additional File: Supplementary File eFigure 1 (A, B) in the attachment for details.
Publication Bias. The funnel plot of the included studies in our final metaanalysis did not suggest a publication bias. The Funnel plot of the included studies in the meta-analysis on the effect of developing CI-AKI after PCI/CAG/CCTA were shown in Additional File: Supplementary File eFigure 2 (A, B).