A combination of metformin and insulin is an effective treatment that benets the cardiovascular and cerebrovascular system in individuals with type 1 diabetes mellitus

Background This study aims to further clarify whether there are cardiovascular and cerebrovascular benets associated with insulin treatment combined with metformin in individuals with type 1 diabetes. Methods Electronic databases were searched for randomized controlled trials in which the ecacy and safety of metformin were compared with those of a placebo for the treatment of cardiovascular and cerebrovascular disease among type 1 diabetes patients, and a meta-analysis was conducted. Results Thirteen cardiovascular studies were identied. In the metformin group, mean carotid intimal media thickness was signicantly reduced by 0.03 mm, ascending aortic pulse wave velocity by 6.3 m/s, descending aortic wall shear stress by 1.77 dyn/cm 2 , insulin daily dose by 0.05 U/kg/d, body weight by 2.27 kg, fat-free mass by 1.32 kg, body mass index by 0.58 kg/m 2 , hip circumference by 0.29 m, and low-density lipoprotein by 0.16 mmol/L. In the metformin group, ow-mediated dilation was increased by 1.29%, glucose infusion rate/insulin by 18.22 µIU/µL, and waist-to-hip ratio by 0.02. The metformin group showed no differences in blood pressure, reactive hyperemia index, waist circumference, triglyceride, total cholesterol, high-density lipoprotein cholesterol, or body mass index Z score. For cerebrovascular studies were identied. But none of them had a risk factor assessment. insulin is another problem and often reaches 1.5-2 U/kg/d, resulting in weight gain, which may be caused by the increase in low-density lipoprotein cholesterol (LDL) or blood pressure (BP) 2 . In adolescents with T1DM, the adolescent surge of hormones and changes in body composition make HbA1c more dicult to control and can lead to insulin resistance (IR) and excessive weight gain; the body mass index (BMI) of T1DM adolescents has been shown to increase faster than that of nondiabetic adolescents 3 . IR also accelerates the progression of atherosclerosis 4 . Atherosclerosis is the basis of cardiovascular disease (CVD), the main cause of death in individuals with T1DM 5 . Even if T1DM patients can maintain a good HbA1c, their risk of cardiovascular morbidity is still three-fold higher than that of the general population 6 . Recent research has found that metformin may improve vascular endothelial dysfunction, which is the rst step in the development of atherosclerosis in patients with T1DM 7 . Metformin may reduce carotid intimal media thickness (cIMT) in patients with T1DM 8 . Even if statins are used for 2 to 4 years, they cannot change cIMT in individuals with T1DM 9 . Studies have shown that atherosclerosis exists in children with T1DM 10 , but vascular functional and structural disorders can be reversed. The effects of adding metformin to the treatment for T1DM on reducing HbA1c have been inconsistent. T1DM is cardiovascular complications, and patients urgently need a program to reduce cardiovascular disease risk. Therefore, we conducted a meta-analysis to further conrm the cardiovascular benets of adding metformin to the treatment plan among patients with T1DM. We analyzed 13 high-quality RCT studies on the effects of cardiovascular risk factors in T1DM patients and found that insulin combined with metformin signicantly reduced the mean and maximum cIMT compared with the effects of insulin alone among T1DM patients. The use of cIMT as a primary outcome is highly consistent with cardiovascular outcomes, with a positive predictive value


Abstract Background
This study aims to further clarify whether there are cardiovascular and cerebrovascular bene ts associated with insulin treatment combined with metformin in individuals with type 1 diabetes.

Methods
Electronic databases were searched for randomized controlled trials in which the e cacy and safety of metformin were compared with those of a placebo for the treatment of cardiovascular and cerebrovascular disease among type 1 diabetes patients, and a meta-analysis was conducted.

Results
Thirteen cardiovascular studies were identi ed. In the metformin group, mean carotid intimal media thickness was signi cantly reduced by 0.03 mm, ascending aortic pulse wave velocity by 6.3 m/s, descending aortic wall shear stress by 1.77 dyn/cm 2 , insulin daily dose by 0.05 U/kg/d, body weight by 2.27 kg, fat-free mass by 1.32 kg, body mass index by 0.58 kg/m 2 , hip circumference by 0.29 m, and low-density lipoprotein by 0.16 mmol/L. In the metformin group, ow-mediated dilation was increased by 1.29%, glucose infusion rate/insulin by 18.22 µIU/µL, and waist-to-hip ratio by 0.02. The metformin group showed no differences in blood pressure, reactive hyperemia index, waist circumference, triglyceride, total cholesterol, high-density lipoprotein cholesterol, or body mass index Z score. For cerebrovascular studies were identi ed. But none of them had a risk factor assessment.

Conclusions
There is strong evidence supporting wider use of metformin to improve cardiovascular and cerebrovascular disease risk among individuals with type 1 diabetes, not just obese or adults.

Background
The incidence of type 1 diabetes mellitus (T1DM) is increasing annually. Notably, a study of 19 countries showed that only 30% of T1DM patients had HbA1c < 7.5% and that at least 30% of T1DM patients had poor HbA1c (> 9.0%) 1 . The complications of diabetes can be reduced by intensive insulin treatment to achieve normal HbA1c. However, the excessive use of insulin is another problem and often reaches 1.5-2 U/kg/d, resulting in weight gain, which may be caused by the increase in low-density lipoprotein cholesterol (LDL) or blood pressure (BP) 2 . In adolescents with T1DM, the adolescent surge of hormones and changes in body composition make HbA1c more di cult to control and can lead to insulin resistance (IR) and excessive weight gain; the body mass index (BMI) of T1DM adolescents has been shown to increase faster than that of nondiabetic adolescents 3 . IR also accelerates the progression of atherosclerosis 4 .
Atherosclerosis is the basis of cardiovascular disease (CVD), the main cause of death in individuals with T1DM 5 . Even if T1DM patients can maintain a good HbA1c, their risk of cardiovascular morbidity is still three-fold higher than that of the general population 6 . Recent research has found that metformin may improve vascular endothelial dysfunction, which is the rst step in the development of atherosclerosis in patients with T1DM 7 . Metformin may reduce carotid intimal media thickness (cIMT) in patients with T1DM 8 . Even if statins are used for 2 to 4 years, they cannot change cIMT in individuals with T1DM 9 . Studies have shown that atherosclerosis exists in children with T1DM 10 , but vascular functional and structural disorders can be reversed. The effects of adding metformin to the treatment for T1DM on reducing HbA1c have been inconsistent.
Diabetes raises the risk of ischaemic stroke through increased atherosclerosis and endothelial dysfunction while also leading to a slower and less complete recovery following stroke 11 . Brain blood vessel dysfunction in diabetic subjects can result in a high level of bleeding following stroke, which worsens prognosis and hinders functional recovery after stroke 12 . Metformin has recently been shown to ameliorate many of the negative effects of stroke by acting through AMPK-mediated pathways 13 . Cerebrovascular diseases that already were classi ed as neurological diseases and no longer belong to CVD after 2018 were separated the statistics and the discussion in this article.
This study aims to further clarify whether the cardiovascular and cerebrovascular system bene ts from the addition of metformin to insulin treatment in individuals with T1DM.

Search strategy and study selection
In this meta-analysis, two of the study authors searched the MEDLINE, EMBASE, and Cochrane Library databases for articles published through 15 July 2020, using the Mesh and text terms "metformin," "diabetes mellitus, type 1" and "carotid intima thickness, atherosclerosis, lipid pro le, vascular function, cardiovascular disease" or "Stroke, transient ischemic attack, cerebral hemorrhage, subarachnoid hemorrhage" to identify randomized clinical trials (RCTs) that compared the e cacy and safety of metformin with those of a placebo for the treatment of atherosclerosis among T1DM patients. We also manually scanned reference lists from trials and review articles to identify other relevant publications 14 .

Data extraction
All potentially eligible scienti c reports assessing the effects of metformin compared to those of a placebo were identi ed by two authors. Each full-text article was reviewed by these two authors 14 . The data were then extracted, and both authors independently assessed the study quality using a standardized approach. Any disagreement in extracted data was settled by other investigators to reach a nal decision. The data extracted included the rst author, year of publication, country where the study was conducted, number of participants, number of patients lost to follow-up, mean age of participants, diabetes duration, intervention duration, clinical outcomes, and safety outcomes. Studies were excluded if they were case reports, letters, comments, or editorials or if no quantitative outcomes were reported 15 . The primary clinical outcomes of interest were cIMT 9 , reactive hyperemia index (RHI) 16 , ow-mediated dilation (FMD) 17 , pulse wave velocity (PWV) 18 , wall shear stress (WSS) 19 , and glucose infusion rate/insulin (M/I) 20 . The secondary outcomes included lipid pro les, which consisted of total cholesterol (TC), triglycerides (TG), high-density lipoprotein (HDL), and LDL; BP; waist and hip circumferences; HbA1c; BMI; homocysteine(HCY) and total insulin daily dose (TIDD). Safety outcomes included severe hypoglycemic events, diabetic ketoacidosis (DKA) events, and gastrointestinal adverse events.

Data analysis
Two authors independently assessed the risk of bias for each included trial using the Cochrane Risk of Bias Tool for Randomized Controlled Trials and the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions 18 . Summary estimates were obtained using a xed-effects model, and a random-effects model was used when I 2 > 50%. We measured the treatment effect by calculating risk ratios (RRs) with 95% con dence intervals (CIs) for dichotomous data and mean differences (MDs) with 95% CIs for continuous data. All statistical analyses were performed using the Cochrane Review statistical software, Review Manager 5.2.

Description of studies
The electronic searches resulted in the identi cation of 378 cardiovascular disease records: MEDLINE 185; EMBASE, 136; and Cochrane, 57. After removing duplicates, 103 records remained, and after examination of the titles and abstracts of these records, we eliminated any studies that clearly did not match our inclusion criteria. We obtained full-text copies of the 33 potentially eligible articles and subjected these to further evaluation. We excluded 9 articles because they were controlled clinical trials, not RCTs. Finally, 24 articles were found to be eligible and included in the meta-analysis ( Table 1. The electronic searches resulted in the identi cation of 0 cerebrovascular disease record. We removed the "type 1 diabetes" Mesh and text word to search. The identi cation of cerebrovascular records was 20: MEDLINE 20; EMBASE, 3; and Cochrane, 1. After removing duplicates, 20 records remained, and after examination of the titles and abstracts of these records, we eliminated any studies that clearly did not match our inclusion criteria. We obtained full-text copies of the 10 potentially eligible articles and subjected these to further evaluation. We excluded 3 articles because they were protocol and 2 articles because metformin was basic drug and 1 article was comment. Finally, 4 articles were found to be eligible and included in the analysis. However, the patients of these 4 articles are not T1DM(Ghotbi2013 44 , Charytan2019 45 , Mellbin2007 46 , Hertog2013 47 ). Details of the studies included are shown in Table 2.  Six studies provided data on BP (heterogeneity: P=0.002; I²=73%). Metformin had no effect on BP (P=0.34). The heterogeneity among the studies disappeared after the removal of Bjornstad 2018 (P=0.12; I²= 46%). The metformin group showed a tendency toward reduced systolic BP (P=0.12). The heterogeneity among the studies disappeared after removing Jacobsen 2009 (P=0.45; I²=0%). Interestingly, the pooled analysis showed that metformin signi cantly reduced the diastolic BP by 2.01 mmHg (95% CI=-2.69--1.34, Z=5.83, P<0.00001) ( Figure 12).
Homocysteine, as a risk factor for stroke, was not mentioned in the 4 articles. These four articles are not systematically analyzed just reviewed. Ghotbi 2013 showed that metformin treatment reduced the rate of nonfatal stroke compared to insulin treatment. Charytan 2019 showed that metformin treatment at baseline was associated with reduced cardiovascular and cerebrovascular complications in patients with diabetes and CKD. Mellbin 2007 showed that the risk for non-fatal myocardial infarction and stroke increased signi cantly in patients on insulin (HR 1.73, 95% CI 1.26-2.37; P= 0.0007), whereas this risk was lower among those on metformin (HR 0.63, 95% CI 0.42-0.95; P= 0.03). Hertog 2013 showed that metformin treatment improved glucose tolerance following cerebral ischaemia in non-diabetic patients, which is a cause of post-stroke damage.
Regarding the incidence of adverse events among patients with T1DM treated with insulin combined with metformin, 6 studies provided data on hypoglycemic events, 5 studies provided data on ketoacidosis, and 6 studies provided data on gastrointestinal adverse reactions, including adverse events experienced by those treated with insulin alone. None of the patients dropped out of treatment due to side effects or the study procedures.

Discussion
The incidence of T1DM complications is as high as 96.5% 48 . The main cause of death among patients with T1DM is cardiovascular complications, and patients urgently need a program to reduce cardiovascular disease risk. Therefore, we conducted a meta-analysis to further con rm the cardiovascular bene ts of adding metformin to the treatment plan among patients with T1DM. We analyzed 13 high-quality RCT studies on the effects of cardiovascular risk factors in T1DM patients and found that insulin combined with metformin signi cantly reduced the mean and maximum cIMT compared with the effects of insulin alone among T1DM patients. The use of cIMT as a primary outcome is highly consistent with cardiovascular outcomes, with a positive predictive value of 96% 49 . This result showed that metformin had a direct effect on preventing the progression of atherosclerosis among T1DM patients, although not by reducing HbA1c because there were no changes in HbA1c in any of the studies.
Another important result was that metformin can signi cantly improve vascular endothelial function in patients with T1DM. Combined insulin and metformin treatment signi cantly increased FMD by 1.29%. FMD is considered a very important standard for the noninvasive evaluation of the vascular endothelium and is widely used in the early detection and prognosis evaluation of a variety of cardiovascular diseases 17 . The treatment of T1DM with insulin combined with metformin signi cantly reduced AA PWV by 6.3 m/s. PWV is an indicator that can re ect the degree of vascular sclerosis, and an increase in PWV re ects endothelial dysfunction 18 . Endothelial dysfunction is the early manifestation of atherosclerotic plaques before angiography or ultrasound and is the initiating factor of atherosclerosis; moreover, the endothelium is the target organ of cardiovascular and cerebrovascular disease risk factors 50 .
The use of metformin can reduce not only LDL but also the use of statins 51 . Some scholars have proposed that in T1DM patients, the risk of CVD is reduced by approximately 20% for every 0.5 mmol/L reduction in LDL 52 . Our meta-analysis showed that metformin reduced LDL by 0.16 mmol/L and reduced the risk of CVD by 6.4%. As seen above, metformin protects vascular endothelial function and reduces LDL level, thus reducing lipid deposition and ultimately reducing cIMT. As a possible mechanism, metformin, via 5'-adenosine monophosphate-activated protein kinase (AMPK) activation, has been thought to directly reduce atherosclerosis and support the molecular effects of statins on cholesterol metabolism 53 . In addition, metformin can inactivate methylglyoxal through the non-AMPK pathway, improve endothelial function, and inhibit the formation of glycosylation end products 54 .
Insulin combined with metformin treatment in patients with T1DM can signi cantly reduce weight by 2.27 kg, BMI by 0.58 kg/m 2 , TIDD by 0.05 U/kg/d, daily insulin by 5.22 U/d, fat-free mass by 1.32 kg, and hip circumference by 0.29 m. In addition, metformin reduced the daily insulin requirement by approximately 16%. These ndings all shows a decrease in central fat. When the data were subanalyzed by sex, only the waist circumference and BMI Z score of women decreased signi cantly 32 . Perhaps appetite suppression may partly explain the signi cant weight loss 27 . Metformin can signi cantly increase the M/I by 18.22 µIU/µL. These results suggest that metformin continues to improve IR, which is an increasingly recognized factor that contributes to the progression of CVD in T1DM patients 4 . Therefore, treatments targeting IR in T1DM patients have become an effective method for the prevention of cardiovascular diseases.
In the past 10 years, there have been 5 meta-analyses on the e cacy of insulin combined with metformin for treating patients with T1DM. The meta-analysis by Al-Khalifa showed that there is no evidence of improving HbA1c but strong evidence of reducing TIDD and BMI. The combination therapy was deemed as safe 55 . The meta-analysis by Liu (2016) showed that the effect may be different between overweight/obese adolescents and nonobese adolescents and can increase the risk of adverse events 56 . Liu's (2015) meta-analysis showed that metformin can reduce TIDD, weight, and lipid levels. Metformin does not increase the incidence of hypoglycemia or ketoacidosis 57 . Vella (2010) concluded that metformin can reduce TIDD 58 . The meta-analysis conducted by Meng showed that insulin combined with metformin could signi cantly reduce TC and LDL in T1DM patients 14 . Those meta-analyses did not include a marker of poor cardiovascular prognosis. The combination of insulin and metformin was associated with a slight increase in gastrointestinal reactions leading to hypoglycemic events in T1DM patients. Therefore, when adding metformin to the T1DM treatment regimen, it is necessary to closely monitor blood glucose and adjust TIDD in a timely manner to reduce the occurrence of hypoglycemia.
Metformin can improve cardiovascular disease risk factors, which are also risk factors for cerebrovascular disease. However, another risk factor for cerebrovascular disease is hyperhomocysteinemia. Zhang 2016 Meta-analysis suggested that metformin could decrease the concentration of Hcy when exogenous B-group vitamins or folic acid supplementation was given 59 . Therefore, metformin can decrease the risk of stroke and has recently been shown to be capable of improving recovery following stroke 60 . Metformin can improve the neurological function and oxidative stress status of acute stroke patients with type 2 diabetes, and its mechanism may be related to the AMPK/mTOR signaling pathway and oxidative stress 61 . The underlying metformin mechanisms include decreasing mitochondrial dysregulation, oxidative stress/reactive oxygen species, blood-brain-barrier breakdown, in ammation of the brain, and neuronal apoptosis 60 . Therefore, there should be more clinical trials to determine whether metformin can be added to improve stroke in T1DM, type 2 diabetes, and Non-diabetic patients, basing on metformin decreasing risk factors of cerebrovascular disease and underlying mechanisms.
Pharmacogenomics must also be considered when patients with T1DM are treated with metformin. The maximum dose of metformin is 2500 mg/d, and a box of metformin (Glucophage) is $4.30, and the annual cost is $315.40. The average duration of the included studies was 8.5 months. Metformin is a costeffective therapy.

Conclusion
Multiple factors proved that metformin can prevent the progression of atherosclerosis among T1DM patients, signi cantly improve endothelial function, reduce lipid deposition, reduce IR, and ultimately reduce cIMT among T1DM patients. There is strong evidence supporting wider use of metformin to improve CVD and cerebrovascular disease risk among T1DM patients, not just among obese or adults.