Literature Search
Based on the selection strategy, we found 2134 citations. Of these, 151 duplicate studies were excluded. After screening the title and abstract, 12 articles remained for assessment of the full text. Four studies[29–32] were excluded for the following reasons: studies without our outcome of interest, study populations did not meet our inclusion criteria, and studies did not report the HR. Ultimately, 8 studies[2, 3, 16–21] fulfilled our inclusion criteria and were included in this meta-analysis (Fig. 1).
Study Characteristics
In aggregate, 8 studies included a total of 1121 patients with diabetes (median age ranging from 52 to 67; 67% were men) who underwent LGE-MRI and whose follow-up ranged from 17 to 70 months. Across the 8 studies, 6 articles[2, 17–21] reported the duration of diabetes, and the mean duration of diabetes was 15 years. A total of 6 studies[2, 3, 16, 19–21] reported the LV ejection fraction, and the mean LV ejection fraction was 57.78%. The presence of LGE was evaluated by visual analysis in 6 publications[2, 3, 18–21]. All of the included studies reported multiple clinical outcomes. The main characteristics of the included articles are shown in Table 1.
First Author, Year
|
Journal
|
patients
|
Hblc, %
|
LGE Defintion
|
DM (type)
|
Mean age (years)
|
Duration of Diabetes (years)
|
LVEF (%)
|
Follow-up duration (months)
|
male
|
LGE(+)
|
Total events
|
Adjusted HR
|
Fibrosis type
|
Type design
|
outcome
|
NOS
|
Bertheau RC,2016
|
Eur Radiol
|
61
|
7.2 (6.5-7.9)
|
visual
|
1 and 2
|
67.5(56.7-71.8)
|
19(14-28)
|
56(46-61)
|
70 (57-72)
|
31
|
17
|
8
|
YES
|
Ischemic
|
Prospective, single-centre
|
MACCE
|
7
|
Heydari B,2016
|
Circ Cardiovasc Imaging
|
173
|
7.9±1.8
|
2 SD
|
NA
|
61.7±11.9
|
NA
|
51.8±17.6
|
34.8±30
|
109
|
88
|
21
|
NO
|
Ischemic
|
Prospective, single-centre
|
MACE
|
7
|
Elliott MD,2019
|
Diabetes Care
|
120
|
NA
|
visual
|
1 and 2
|
52±13
|
17±11
|
63±9
|
46 (33-64)
|
65
|
23
|
19
|
YES
|
Ischemic
|
Prospective, two-centre
|
MACE
|
9
|
Yoon YE,2013
|
Eur Radiol
|
120
|
7.4±1.5
|
visual
|
2
|
67±9
|
11±11
|
63±6
|
27 (7-112)
|
83
|
18
|
10
|
NO
|
Ischemic
|
Retrospective, single-centre
|
MACE
|
7
|
Giusca S,2016
|
Eur Heart J Cardiovasc Imaging
|
328
|
NA
|
visual
|
NA
|
67±11
|
NA
|
57.7±11.6
|
35 (23-51.6)
|
250
|
176
|
26
|
YES
|
Ischemic and nonischemic
|
Prospective, multicentre
|
MACE
|
8
|
Bamberg F,2013
|
Radiology
|
61
|
7.2 (6.5-7.9)
|
visual
|
1 and 2
|
67.5(56.7-71.8)
|
19(14-28)
|
56(46-61)
|
70 (57-72)
|
31
|
17
|
18
|
YES
|
Ischemic
|
Prospective, single-centre
|
MACCE
|
7
|
Kwong RY,2008
|
Circulation
|
107
|
7.3±1.6
|
2 SD
|
NA
|
59±12
|
10.7±8.5
|
NA
|
17 (6-57)
|
67
|
30
|
38
|
YES
|
Ischemic and nonischemic
|
Prospective, single-centre
|
MACCE
|
9
|
Yoon YE,2012
|
Radiology
|
151
|
7.4±1.6
|
visual
|
NA
|
67±9
|
14±11
|
NA
|
30(6-103)
|
113
|
58
|
24
|
NO
|
Ischemic
|
Retrospective, single-centre
|
MACE
|
6
|
Columns represent n(%) or mean±SD or median (IQR); DM, diabetes mellitus; LGE, late gadolinium enhancement; LVEF, left ventricular ejection fraction; NOS, Newcastle-Ottawa Scale; HR, hazard ratio; NR, not reported; MACCE, major adverse cardiac and cerebrovascular events; MACE, major adverse cardiac events.
|
Table 1
Description of the Studies Included in the Meta-Analysis
Among the 8 selected studies, 6 studies[16–21] (75%) were conducted in a single center (Germany, n = 2; USA, n = 2; Japan, n = 2), and 2 studies[2, 3] were performed in multiple centers (USA, n = 1; Europe, n = 1). Five articles[2, 3, 17, 20, 21] (62.5%) reported adjusted HR. Six studies[2, 16, 18–21] reported patients with ischemic fibrosis, and the remaining 2 studies[3, 17] reported patients with ischemic and nonischemic fibrosis.
Of the 8 eligible studies, 7 received 7 to 9 scores, and the mean NOS score was 7.5. Overall, the aforementioned analysis showed that the included articles had high quality (Table 1). Among the identified studies, there was no risk of publication bias assessed by visual analysis of the funnel plot (Fig. 2).
Prevalence of LGE and AERs
Across the 8 studies, the prevalence of myocardial fibrosis detected by LGE ranged from 15% to 62%, and the prevalence of LGE in the total sample was 38.09% (n = 427). Furthermore, a total of 164 events occurred in the diabetes group (n = 1121) during the median follow-up of 3.4 years. Patients with diabetes had AERs for MACCE of 4.3%. However, only 3 studies[2, 19, 21] reported a total of 301 patients with diabetes. Among these patients, 19.27% (n = 58) had myocardial fibrosis detected by LGE, with 27 events occurring over a median follow-up of 3.9 years. The AERs of patients with diabetes and LGE was 11.94%.
MACCE and MACE
A total of 8 studies reported the outcome of MACCE or MACE, and the presence of myocardial fibrosis detected by LGE was a strong predictor of MACCE and MACE in patients with diabetes (random-effects HR 3.87, 95% CI 2.58–5.80; P = 0.000) (Fig. 3). Low heterogeneity (I2 = 15.1%, P = 0.311) existed in the meta-analysis. In addition, sensitivity analysis performed by excluding 1 study each time found that the HR values were not significantly changed.
In the analysis of the outcome of MACCE, 3 articles[17, 20, 21] were included in this subgroup meta-analysis, including 64 participants with LGE and 165 diabetes without LGE, with a total of 64 MACCE outcomes during the follow-up period. Myocardial fibrosis detected by LGE was associated with an increased risk of MACCE in patients with diabetes. The pooled random-effects HR was 2.58 (95% CI 1.42–4.71; P = 0.002), with no evidence of heterogeneity (I2 = 14.1%; P = 0.312) (Fig. 3).
To explore the association between myocardial fibrosis and the outcome of MACE in patients with diabetes, we included 5 articles[2, 3, 16, 18, 19] that provided a subgroup meta-analysis. The results showed that the presence of LGE in diabetes was associated with a significantly higher risk of MACE. As in the discovery analyses, the pooled HR was 5.28 (95% CI 3.20–8.70; P = 0.000) with no significant heterogeneity (I2 = 0%; P = 0.643) from random effects (Fig. 3).
To further verify the robustness of the results, we grouped all included studies by adjusted or non-adjusted HR. In patients with diabetes, myocardial fibrosis detected by LGE was associated with an increased risk of MACCE and MACE in a subgroup meta-analysis with or without adjusted HR. The pooled HRs were 3.52 (random-effects, 95% CI 2.02–6.16; I2 = 35.8%) and 4.63 (random-effects, 95% CI 2.35–9.14; I2 = 0%), respectively. There was no significant heterogeneity among the studies (Fig. 4).
To evaluate the pattern of myocardial fibrosis effects, we further calculated a pooled HR by source of diabetes with different patterns of myocardial fibrosis. In patients with diabetes, ischemic fibrosis detected by LGE was significantly associated with increased MACCE and MACE (random-effects HR 3.75, 95% CI 2.11–6.69; I2 = 38.3%). Furthermore, all myocardial fibrosis detected by LGE in patients with diabetes may increase the risk of MACCE and MACE (random-effects HR 4.27, 95% CI 2.17–8.37; I2 = 0%) (Fig. 5).
To confirm whether there were similar results in patients with preserved LV ejection fraction, we conducted a subgroup meta-analysis with 6 studies. Among individuals with diabetes and LV ejection fraction > 50%, the presence of myocardial fibrosis assessed by LGE was significantly associated with MACCE and MACE. The pooled HR was 3.98 (95% CI 2.22–7.25; P = 0.000) with random effects, and there was medium heterogeneity among the studies (I2 = 37.9%; P = 0.153) (Fig. 6).