Cognitive protection of incretin‐based therapies in patients with type 2 diabetes mellitus: A systematic review and meta‐analysis based on clinical studies

Abstract Aims/introduction Cognitive dysfunction, including mild cognitive impairment and dementia, is increasingly recognized as an important complication of type 2 diabetes mellitus. The aims of the preset study was to investigate the cognitive protection of incretin‐based therapies, including glucagon‐like peptide‐1 receptor agonists and dipeptidyl peptidase‐4 inhibitors, in patients with type 2 diabetes mellitus. Materials and Methods PubMed, EMBASE, Cochrane library, Web of Science and PsycINFO were searched from the inception through 17 January 2023 for randomized controlled trials and cohort studies on the association between incretin‐based therapies and cognitive function. A total of 15 studies were finally included in our systematic review, and eight of which were incorporated into our meta‐analysis. Results Pooled results showed that the Mini‐Mental State Examination score in incretin‐based therapy groups was increased by 1.20 compared with the control group (weighted mean difference 1.20, 95% confidence interval 0.39–2.01). The results of eight studies assessed by the Newcastle Ottawa Quality Assessment Scale and the Cochrane Collaboration's tool, and the quality of the eight studies were at a relatively high level. Egger's regression did not show significant publication bias. Conclusions Current evidence shows that incretin‐based therapies might be more effective, when compared with the other hypoglycemic drugs, for cognitive improvement in patients with type 2 diabetes mellitus.


INTRODUCTION
In the past decades, the prevalence of diabetes has risen rapidly and affected the quality of life of patients 1 . Cognitive dysfunction, including mild cognitive impairment and dementia, is increasingly recognized as an important complication of type 2 diabetes mellitus 2 . Approximately 10% of dementia patients have diabetes 3 . Compared with patients without diabetes, type 2 diabetes mellitus patients have a higher risk of cognitive impairment 4,5 , especially when accompanied by renal 6 and cardiovascular complications 7 . The English Longitudinal Study of Aging showed that cognitive decline was associated with prediabetes and diabetes, and the increment in hemoglobin A1c was significantly associated with an increased rate of decline in cognitive function 8 .
Incretin-based therapies include glucagon-like peptide-1 receptor agonists (GLP-1 RAs) and dipeptidyl peptidase-4 (DPP-4) inhibitors 9,10 . Glucagon-like peptide-1 (GLP-1) is an endogenous peptide hormone released by intestinal L-cells in response to a meal. GLP-1 stimulates insulin secretion from the pancreatic b-cells under hyperglycemic conditions and reduces glucagon secretion from the a-cells, recovering insulin sensitivity and enhancing glycemic homeostasis 11 . The effect of DPP-4 inhibitor is indirectly mediated by GLP-1. GLP-1 is mainly metabolized through DPP-4 enzyme degradation, which is expressed in many organs, such as the liver, pancreas, intestine and brain 12,13 . Incretin-based therapies are increasingly used in type 2 diabetes mellitus, and have also emerged as a potential therapeutic agent for Alzheimer's disease (AD) 14 , as well as vascular brain injury. Meta-analysis based on randomized controlled trials shows that GLP-1 RAs can significantly reduce the risk of stroke in patients with established atherosclerotic cardiovascular disease 15,16 . Although some studies examined the association between incretin and cognitive function in type 2 diabetes mellitus, their findings were inconsistent [17][18][19][20][21][22][23][24][25][26][27] . Therefore, we carried out a systematic review and meta-analysis to investigate the association between incretin-based therapies and cognitive function in patients with type 2 diabetes mellitus.

MATERIALS AND METHODS
The present study was carried out according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement 28 . This study was registered on the International Prospective Register of Systematic Reviews (PROSPERO), and the registration number is CRD42020159750.

Search strategy and literature search
We carried out literature searches from online databases of literature including: PubMed, EMBASE, Cochrane Library, Web of Science and PsycINFO. The search date ranges from the inception of each database through 17 January 2023. We used "Glucagon-Like Peptide-1 Receptor" and "Dipeptidyl-Peptidase IV Inhibitors" as keywords or MeSH/EMTREE terms, accompanied with other relevant free words to search these databases; details of the search strategies are provided in Table S1.
Literature screening and selection Inclusion criteria included: (1) randomized controlled trials (RCTs) or cohort studies involving DPP-4 inhibitors or GLP-1 RAs compared with placebo or other antidiabetic agents (metformin, insulin, sulfonylureas, thiazolidinediones, alphaglucosidase inhibitors, sodium-glucose cotransporter 2 inhibitor) in patients with type 2 diabetes mellitus were included in our analysis (Table 1) were included in the present meta-analysis. The eligibility of studies was assessed independently by three reviewers (SQY, SBC and FS), with any disagreement being resolved by consensus.

Data extraction
Data from studies selected for full-text review were extracted in duplicate for quality assessment and data analysis. Data were extracted by using Excel (Microsoft, Redmond, WA, USA), including study information (author, publication year, sample size, trial duration, types of intervention and control), population characteristics (background therapy, diabetes duration, age, baseline level of hemoglobin A1c), reported outcomes of cognitive function and information of methodology.

Quality and risk of bias assessment
We assessed the quality and risk of bias of these included studies by using appropriate tools based on the type of studies. The Newcastle Ottawa Quality Assessment Scale and the Cochrane Collaboration's tool were used for evaluating cohort studies 29 and RCTs 30 , respectively. The Grading of Recommendations Assessment, Development and Evaluation (GRADE) framework was also used to assess the quality of evidence for cognitive function, which assessed the within-study limitations, imprecision, heterogeneity, indirectness and publication bias for the MMSE outcome. The whole assessment process was completed independently by FQL and SBC in parallel, and FS confirmed the final extraction when there was any inconsistency.

Statistical analysis
After screening and data extraction, eligible data were pooled through meta-analysis carried out using Review Manager (5.3 version; Cochrane Collaboration, London, UK). The results of the comparison of cognitive function between the intervention and control group are shown as the mean difference (MD) of the MMSE score and its 95% confidence interval (95% CI). The heterogeneity of these studies was measured by the I 2statistic. Egger's regression was carried out to test publication bias because of small-study effects. We carried out a sensitivity analysis including RCTs and cohort studies. We did not carry out further subgroup analysis or meta-regression to explore potential sources of heterogeneity given the insufficient number of included studies.

Literature search
Our search strategy resulted in the identification of 7,172 articles ( Figure 1). After the extensive review of the titles and abstracts of these articles, 43 articles were identified for full-text review. A total of 15 studies were finally included in our systematic review, and eight of them were incorporated into our meta-analysis.
One of the three RCTs showed positive results (WMD 4.08, 95% CI 3.39-4.77) 22 and the other two showed negative results 20,26 . Three of the four cohort studies showed positive results, and one showed a negative result. Of the four studies evaluated by other cognitive tools, two studies suggested that the use of incretin drugs could improve the cognition of patients with type 2 diabetes mellitus, and two studies showed no improvement in cognition performance or cognition function. The remaining three studies without MMSE baseline or specific values suggested that DPP-4 inhibitors can significantly improve cognitive function in patients with type 2 diabetes mellitus compared with other oral hypoglycemic agents.
Subgroup analysis of sulfonylureas and other oral hypoglycemic drugs Meta-analysis was carried out for the study of sulfonylureas as the control group and the study of non-sulfonylureas as the control group, and the results were WMD 1.65, 95% CI -0.80 to 4.09 and WMD 0.93, 95% CI -0.09 to 1.94, respectively. Full-text articles excluded, with reasons (n=28) Articles excluded based on the reasons: The stratification result is consistent with the general conclusion. See Figure S2.

Quality and risk of bias assessment results
The results of eight studies assessed by the Newcastle Ottawa Quality Assessment Scale and the Cochrane Collaboration's tool are shown in Tables S2 and S3. The quality of the eight studies included in the meta-analysis were at a relatively high level. Egger's regression did not show significant publication bias. Publication bias refers to the research results with statistical significance being more likely to be reported and published than the results without statistical significance and that are invalid. According to GRADE, the quality of evidence was rated as high for the cognitive function outcome in the comparison between incretin-based therapies and control groups ( Table 2).

DISCUSSION
The present meta-analysis showed that incretin-based therapies improve cognitive function (evaluated using MMSE) in patients with type 2 diabetes mellitus. In our review, the significant association between incretin-based therapies and cognitive function was observed in RCTs and observational studies. The overall quality of the studies included in the meta-analysis is high. The MMSE is a cognitive function assessment with 30 questions, involving six aspects: orientation, registration, attention and calculation, memory, language and visual spatial ability 35     tool for cognitive assessment in clinical and scientific research 36,37 . The score is calculated based on the scores of correct items, and the score of missing items is 0 38 . Through verification, it is suggested that there is a high correlation between MMSE scores and other cognitive indicators and the scores of activities of daily living. It suggests that there is a high correlation between MMSE and other cognitive indicators, as well as activities of daily living 39 . In addition to the common microvascular and macrovascular complications, cognitive decline is considered as an emerging consequence of type 2 diabetes mellitus 40 . Diabetes is a major risk factor for AD, vascular dementia and Parkinson's disease (PD). These neurodegenerative diseases are closely related to impaired glucose metabolism in diabetes. Among the eight trials included in the present meta-analysis, the results of six trials 17,18,[21][22][23]27 suggested that incretin drugs have the effect of improving the cognition of diabetes patients, and the results of two trials 20,26 were negative. Three trials suggested that the blood glucose of the incretin group was significantly lower than that of the control group 17,20,23 , and four trials suggested that there was no significant difference between the two groups 18,21,26,27 . In terms of hypoglycemic events, all results suggested that the risk of hypoglycemia in the incretin group was significantly reduced. Therefore, controlling blood glucose to reach the standard and avoiding hypoglycemic events might improve the cognitive function of diabetes patients.
Insulin resistance and brain insulin signal transduction defects can accelerate the onset of neurodegenerative diseases by reducing brain metabolism. Although an association between diabetes and cognitive dysfunction has been observed, the effect of hypoglycemic drugs on cognitive dysfunction is still unclear. GLP-1, as a physiological regulator of the central nervous system, enhances learning and memory functions by restoring insulin signal transduction 41 . Increasing evidence shows that GLP-1 analogs have many extrapancreatic effects. GLP-1 can affect cellular signal transduction through the blood-brain barrier, such as neuroinflammation and mitochondrial function in the central nervous system 42 . An open trial of exenatide in the treatment of moderate PD showed continuous improvement in motor and cognitive functions of patients treated with exenatide 43 . A double-blind trial using exenatide weekly preparation showed that exenatide had a positive effect on off-medication motor scores in patients with moderate PD 14 . Dumbrill and Moulton 44 carried out a systematic review examining the effects of incretin-based therapies on neurocognitive outcomes in human studies. The results showed that there was evidence to support that DPP-4 inhibitor can improve cognitive function of patients, whereas GLP-1 can improve cerebral glucose metabolism 44 . In a randomized, placebo-controlled, double-blind study, liraglutide improved glucose metabolism and cognitive function in AD patients 45 . The results of clinical research on incretinbased therapies are consistent with those of preclinical research. Xue et al. 22 found that treatment with DPP-4 inhibitor for 6 months significantly improved the cognitive ability and increased the Ab1-42/Ab1-40 value in elderly patients with type 2 diabetes mellitus combined with post-stroke mild cognitive impairment. These results suggested that DPP-4 inhibitor could improve the cognitive function in patients with type 2 diabetes mellitus, which might be associated with the improvement of Ab gathering.
In animal models, incretin-based therapies can improve the pathological changes of AD. Some studies have shown that GLP-1 can reduce oxidative stress and inflammatory reaction of neurons. The study of a PD mouse model found that GLP-1 can maintain synaptic plasticity, thereby improving the motor and non-motor deficits 42,46 . In the diabetes rat model of cerebral ischemia/reperfusion injury, recombinant GLP-1 improves neural injury by inhibiting oxidative stress response 47 . Using streptozotocin to prepare animal models of diabetes and cognitive decline, it was found that liraglutide could improve the hippocampal neurodegeneration of animals 48 . In mice models of AD, it was found that the treatment of sitagliptin and linagliptin could increase the level of GLP-1 in the brain, thereby reducing memory impairment, tau phosphorylation and neuroinflammatory reaction 49,50 . A study on AD transgenic mice found that intraperitoneal injection of exenatide can reduce the pathological changes in the hippocampus of mice, thereby improving cognitive function 51 . In the rat model of PD, vildagliptin plays a variety of neuroprotective effects, thereby improving the cognitive function of rats 52 .
Compared with Luan et al. 53 and Jin et al. 54 , the novelty of this meta-analysis includes the following three aspects. First, our research includes GLP-1 RA S and DPP-4 inhibitors, and the drug is more comprehensive. Second, 15 studies were included in this meta-analysis, including four RCTs. The quantity and quality of the studies were better than those of the aforementioned two studies. Third, in addition to the MMSE and Montreal Cognitive Assessment, the indicators of cognitive evaluation include the Trail Making Test, Verbal Fluency Test, Babcock Story Recall Test, Babcock Story Recall Test and Attention Matrices Test. The indicators of cognitive function evaluation are more comprehensive. Therefore, the results of the present study on the cognitive function of incretin in diabetes patients are more objective than those of the aforementioned two studies, which provides a basis for more clinical studies on the cognitive function of such drugs and diabetes patients.
There were several limitations to the present review. First, the overall quality of the evidence was assessed as moderate in the GRADE framework, so current evidence suggests an association of incretin-based therapies with cognitive function, but cannot confirm the causal effects of the therapies on cognitive function. Second, there was substantial heterogeneity in the effect estimates. Different study types might be one of the possible explanations. Other possible reasons for heterogeneity could include the differences in study population, treatment duration, length of follow up and methodological quality. Due to the limited number of eligible studies, however, we could not carry out further subgroup analysis or meta-regression to explore these potential sources of heterogeneity. Third, our study was limited to studies published in English only. Fourth, due to the small number of studies finally included, there will inevitably be some publication bias. The result of the present meta-analysis showed that incretinbased therapies are associated with the reduced risk of cognitive impairment in patients with type 2 diabetes mellitus. Due to the limited number of RCTs included in this study and the unstable results, it is necessary to refer to the results of this study with caution to guide clinical practice. However, the conclusion of RCTs in this study is basically consistent with that of prospective cohort studies, which also increases the confidence of using this evidence. Further high-quality studies are required to confirm the present findings.

SUPPORTING INFORMATION
Additional supporting information may be found online in the Supporting Information section at the end of the article.