Specular microscopy findings in diabetic patients: a systematic review and meta-analysis Authors: Ali Sharifi, Amir Masoud Moeini, Mohsen Nabi-Afjadi Published: 2024 ABSTRACT AbstractObjectiveTo evaluate the specular microscopy findings in patients with diabetes mellitus (DM) through a systematic review and meta-analysis.MethodsA systematic search with the appropriate keyword was done in the online databases including PubMed, Scopus, and Web of Sciences. The records were imported into EndNote software and duplicated records were removed. The title, abstract, and full text of the articles were evaluated carefully. STATA software also was used for all data analyses. The mean difference (MD) and 95% confidence intervals (95% CIs) were used to examine the data.ResultsAfter a systematic search, 781 records were evaluated and finally, 14 of them were selected for final analysis. The results showed that the mean differences in average cell size (MD = 21.22, 95% CI 10.43, 32.01, P 0.001), central corneal thickness (MD = 9.90, 95% CI 6.07, 13.73, P 0.001), and coefficient of variation (MD = 1.11, 95% CI 0.11, 2.11, P = 0.03) were higher in diabetic patients in comparison with healthy cases. However, the mean differences in endothelial cell density (MD = -114.29, 95% CI -158.34, -70.24, P 0.001) and hexagonal cell ratio (MD = -3.52, 95% CI -5.79, -1.25, P < 0.001) were lower in diabetic cases.ConclusionsPatients with DM have higher average cell size, corneal thickness, and coefficient of variation. On the other hand, the endothelial cell density and hexagonal cell ratio were reduced in diabetic cases.Keywords: Cornea, Specular microscopy, Diabetes Mellitus, Retinopathy Introduction Diabetes mellitus (DM), comprising both type 1 and type 2 variants, is a chronic metabolic condition characterized by impaired insulin production, insulin function, or both, leading to hyperglycemia and associated micro/macrovascular complications [1, 2]. Over the past few decades, there has been a dramatic rise in the prevalence of DM, attributed primarily to factors such as obesity, aging populations, and sedentary lifestyles. In recent years, DM has even been dubbed the “Black Death” due to its widespread impact [2, 3]. A well-known risk factor for visual impairment in DM patients is diabetic retinopathy. The cornea and retina are most severely affected by DM’s effects on the eyes [4, 5]. 70% of DM patients suffer from corneal problems. This is known as diabetic keratopathy including recurrent erosions, delayed wound healing, ulcers, and edema [5, 6]. Prior studies have demonstrated that corneal endothelial cells in DM patients exhibit morphological abnormalities, including an increase in polymegathism, pleomorphism, and central corneal thickness (CCT), as well as a decrease in endothelial cell density (ECD) and hexagonality [7–9]. These alterations could be attributed to long-term metabolic changes brought on by hyperglycemia at the cellular level, primarily affecting the single layer of coherent endothelial cells, while impaired endothelial function in the cornea can upset the equilibrium of stromal hydration, resulting in corneal edema, changes in corneal transparency, and decreased corneal sensitivity in individuals with DM [10]. These morphological and functional alterations in the cornea are thought to be caused by diabetics having less Na+- K + ATPase activity than healthy individuals, and as a result, the cornea compensates for intraocular pressure by suffering damage [6, 11]. Although individuals with DM have been linked to altered corneal endothelium morphology in the past, there is little agreement among investigations. This study’s objective was to analyze the results of specular microscopy in diabetes patients using a systematic review and meta-analysis. Methods Literature search strategy This research involved a thorough review and meta-analysis. The systematic review was carried out by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) standard. The PRISMA flow chart for this review is shown in Fig. 1. This study has been registered in the Kerman University of Medical Sciences local research ethics committee (IR.ARI.KUI.REC. 1401.223). The PRISMA flow chart In May 2023, a thorough search was conducted using the most relevant keywords after consulting with specialists in the fields of endocrine and ocular illnesses. For this study’s database searches, PubMed, Scopus, and Web of Sciences were all used. We incorporated MeSH entries and other thesaurus terms regarding two components of the population (patients with Diabetes mellitus OR Diabetic Retinopathy) AND (Specular microscopy OR Cornea). In addition, we examined the reference list of all included articles to identify other series published previously that were not identified in our primary search. Inclusion and exclusion criteria Duplicate results were eliminated using the EndNote X8 program (V8.0.1, Clarivate Analytics), which was used to import the data. All published articles had been conducted on humans and had investigated specular microscopy findings in patients with DM compare with healthy control were included in the systematic review. Results obtained from opinion pieces, review articles, letters, case reports, book chapters and websites were excluded from our study. No restrictions for language of publication or geographic location were applied. We chose all observational studies (including cross-sectional, case-control and cohort study) comparing the alterations in the cornea of diabetes patients with those of healthy controls. Study selection and appraisal Three steps were taken to analyze the outcomes. The authors reviewed article titles in the first phase, eliminating any extraneous content. The remaining publications’ abstracts were then examined by the writers, and they eliminated any extraneous material. The writers attempted to obtain the whole texts of the remaining articles in the third phase to analyze them. To get the indicated complete texts for any remaining papers for which the full texts were not yet accessible, the authors emailed the related and corresponding authors. Data extraction from eligible articles was performed independently by two reviewers (AS and AM). To reduce selection bias, each screening step was carried out independently. Discrepancies were resolved through discussion, and where consensus was challenging, opinion of a third expert reviewer (MN) was sought for resolution. Finally, the writers held a discussion and decided which papers would be chosen as the final submissions. Data analysis The National Institutes of Health quality evaluation method also assessed the potential of bias. The risk of bias assessment figure was also drawn. The mean difference (MD) and 95% confidence intervals (95% CIs) were used to examine the data. All data analyses were performed using STATA version 17 (StataCorp, 2020. StataCorp LLC, College Station, TX). When the studies’ clinical and statistical heterogeneity limits were met, a meta-analysis was performed. Utilizing the Cochran Q test and I2 statistics, heterogeneity was measured. According to the Cochrane protocol, a statistically significant heterogeneity that justifies the use of random effect and otherwise fixed effect has a P value of 0.1 for Chi-square testing of the Q statistic or an I2 > 50%. Results Study characteristics A total of 781 records were considered in this evaluation; after duplicate data were removed, 396 articles were left. 221 of the films were eliminated after screening because of their unrelated material. After that, 164 out of 185 abstracts that were examined were disqualified. 21 articles were ultimately chosen for full-text analysis. 14 studies were chosen for the final analysis after a thorough examination, and their features are shown in Table 1 [9, 12–24]. Table 1The summary of the studiesAuthorYearSample sizeGroupsEndothelial Cell Density (cells/m2)Average cell size (µm2)Hexagonal cell ratio (%)Coefficient of Variation (%)Central corneal thickness (µm2)Jha2021592DM2484.5 ± 299.5-39.9 ± 5.240.3 ± 6.1522.1 ± 36.6596Control2555.9 ± 258.2-44.6 ± 6.037.2 ± 6.1514.9 ± 37.1Chowdhury2021131DM2762.08 ± 320.21-55.34 ± 5.8231.60 ± 3.51514.54 ± 38.17131Control2771.46 ± 336.7-56.94 ± 7.0030.85 ± 6.68511.65 ± 32.64Eichwalder202150DM2263 ± 352-57 ± 2132 ± 9532 ± 3950Control2267 ± 292-55 ± 2233 ± 9515 ± 35Tasli2020195DM2427.94 ± 427.10419.69 ± 71.9866.94 ± 7.1732.37 ± 6.16539.01 ± 41.19100Control2615.56 ± 106.80374.74 ± 14.1178.48 ± 6.5630.29 ± 4.08525.90 ± 28.79Papadakou202072DM2297.9 ± 311.3415.6 ± 61.346.3 ± 8.2-543.2 ± 38.788Control2518.3 ± 243.7401.2 ± 43.248.8 ± 8.3-531.9 ± 37.2Alagorie2020120DM2295 ± 311411.6 ± 52.149.1 ± 10.236.0 ± 6.2544.2 ± 38.0112Control2501 ± 302408.2 ± 48.855.2 ± 10.034.9 ± 4.0521.5 ± 31.5Islam2017149DM2494.47 ± 394.10415.31 ± 95.3452.42 ± 6.9436.03 ± 4.26-149Control2574.46 ± 279.97398.99 ± 50.7852.83 ± 7.4435.86 ± 4.39-El-Agamy201757DM2,491.98 ± 261.08-33.24 ± 10.25-545.61 ± 30.3945Control2,629.68 ± 293.45-34.24 ± 8.73-539.42 ± 29.22Storr-Paulsen2014107DM2578 ± 77-58.0 ± 1.431.9 ± 1.0546 ± 7128Control2605 ± 66-58.5 ± 1.131.9 ± 0.9538 ± 5Choo2010100DM2541.6 ± 516.4397.6 ± 84.341.1 ± 19.667.2 ± 47.2517.3 ± 53.4100Control2660.1 ± 515.5380.0 ± 80.445.2 ± 20.658.2 ± 43.0510.8 ± 71.9Keoleian199214DM2883 ± 280-68 ± 8--14Control3131 ± 483-74 ± 8--Shetlar198919DM-426 ± 66---19Control-395 ± 36---Schultz198414DM2472 ± 49.2406.6 ± 7.7---10Control2625 ± 54382.4 ± 7.9---Busted198134DM2686 ± 356----47Control2853 ± 386---- The summary of the studies The result of the Cochrane Collaboration’s tool for assessing risk of bias in included randomized trials shows a low bias as illustrated in Fig. 2. Risk of bias in included studies Average cell size A meta-analysis of 7 studies with 669 cases and 578 controls examined average cell size. Because the investigation revealed heterogeneity (I2 = 72.6%), a random-effects model was used. According to the analytical findings, there was a statistically significant difference between the average cell size in diabetic patients and the healthy population (MD = 21.22, 95% CI 10.43, 32.01, P 0.001) (Fig. 3). The forest plot of average cell size Central corneal thickness Heterogeneity was found in the study (I2 = 59.69%), so a random-effects model was applied. According to the analytical results, the central corneal thickness in patients with diabetes was greater than in the healthy population, which was of statistical difference (MD = 9.90, 95% CI 6.07, 13.73, P 0.001) (Fig. 4). The forest plot of corneal thickness Coefficient of variation A coefficient of variation analysis was done on 8 studies involving 1444 diabetic cases and 1366 healthy cases. I2 was calculated as 86.41% hence, a random-effects model was applied. The results showed that the coefficient of variation was higher in patients with diabetes (MD = 1.11, 95% CI 0.11, 2.11, P = 0.03) (Fig. 5). The forest plot of coefficient of variation Endothelial cell density For the examination of endothelial cell density, 13 studies were included, encompassing 1635 diabetes individuals and 1570 healthy cases. I2 was estimated to be 83.49%; hence, a random-effects model was used. In diabetic individuals, endothelial cell density was considerably reduced (MD = -114.29, 95% CI -158.34, -70.24, P 0.001) (Fig. 6). The forest plot of endothelial cell density Hexagonal cell ratio Meta-analysis results on hexagonal cell ratio showed that 11 studies included with 1587 cases and 1513 controls. As I2 was 96.71%, a random effect was used for analysis. Results showed that the hexagonal cell ratio was lower in diabetic patients in comparison with healthy cases (MD = -3.52, 95% CI -5.79, -1.25, P < 0.001) (Fig. 7). The forest plot of hexagonal cell ratio Publication bias The publication bias was evaluated with Egger’s test for the abovementioned parameters. Results showed no apparent publication bias in the studied parameters (All P were more than 0.05). The result of publication bias estimation is shown in Table 2. Table 2The results of Egger’s testParametersEgger’ testPublication biasAverage cell size0.8919NoCentral corneal thickness0.4611NoCoefficient of variation0.5536NoEndothelial Cell Density0.2382NoHexagonal cell ratio0.7460No The results of Egger’s test Discussion The results of the specular microscopy in diabetic patients were reviewed in this systematic review and meta-analysis. Relevant articles were chosen for data extraction after completing a methodical search. The metrics of endothelial cell density, average cell size, hexagonal cell ratio, coefficient of variation, and central corneal thickness were examined after 14 publications had undergone comprehensive examination. The earliest publication that examined corneal thickness and the corneal endothelium in diabetes mellitus was the Busted et al. investigation. They studied two patient groups, including diabetes instances and individuals who were in good condition. In diabetes instances, endothelial cell density was considerably reduced [24]. Schultz et al. investigated diabetic endothelial alterations. In contrast to an age-matched non-diabetic population, their findings showed that the corneal endothelium in type II diabetes displayed a significantly higher coefficient of variation, a decrease in the percentage of hexagonal cells, and a low figure coefficient [23]. In the corneas of diabetes donors and age-matched nondiabetic controls, Shetlar et al. examined the thickness and makeup of Descemet’s membrane. They could not find any quantitative or qualitative differences between the two groups that were statistically significant. Shetlar’s study used a sample size of 19 patients [22]. In the Keoleian et al. investigation, diabetes individuals’ endothelium was compared using specular microscopy. Diabetes patients’ eyes displayed higher corneal autofluorescence, higher intraocular pressure, a lower percentage of hexagonal endothelial cells, and a higher coefficient of variation of endothelial cell area [21]. Choo et al. evaluated the central corneal thickness and corneal endothelium structure between type II diabetic patients and control individuals who did not have diabetes. They discovered that diabetes subjects have lower endothelial cell density [20]. In type II diabetic and non-diabetic individuals, Storr-Paulsen studied the density and shape of ocular endothelial cells. They discovered that in people with adequate glycemic control, type II diabetes did not affect corneal cell density or shape. They also looked at HbA1c levels and discovered that it was linked to a reduction in endothelial cell density. The diabetic group had considerably thicker central corneas [19]. El-Agamy et al. used an EM-3000 specular microscope to evaluate the corneal endothelial cell density, morphological characteristics, and central corneal thickness in type 2 diabetes mellitus with age-matched, non-diabetic control patients. Their research revealed that type II diabetes significantly lowers endothelial cell density. In diabetic instances, the central corneal thickness rose and the proportion of hexagonal cells decreased, although these changes lacked statistical significance [9]. To assess the link between these measures and the length of diabetes mellitus, glycemic status, and severity of diabetic retinopathy, Islam et al. evaluated corneal morphological parameters between diabetics and age-matched non-diabetic control patients. In this investigation, it was shown that those with diabetes had considerably decreased mean corneal endothelial cell density [18]. In individuals with diabetes mellitus, Papadakou et al. evaluated the features of corneal endothelial cells and central corneal thickness. They discovered that there was a reduction in endothelial cell density in the diabetic group, suggesting that diabetes may alter the corneal endothelium. Additionally, changes in endothelial cell density were substantially correlated with HbA1c levels and the severity of retinopathy. Due to the heterogeneity of the HbA1c values, the current meta-analysis did not examine them [16]. Tasli et al. sought to ascertain how the morphological properties of the cornea correlated with the traits and laboratory findings of diabetes patients. They concluded that keratopathy, which has an increased coefficient of variation, reduced endothelial cell density, and proportion of hexagonal cells, is a serious consequence of type 2 diabetes. Changes in corneal findings grew together with the progression of diabetes [15]. The corneal endothelium morphology and cell density of diabetic smokers and nonsmokers were similarly investigated by Eichwalder et al. The findings showed that diabetes and smoking both significantly reduce endothelial cell density [13]. In research, Chowdhury et al. looked at the thickness and shape of the corneal endothelium layer in people with diabetes mellitus. Patients with diabetes mellitus were found to have larger central corneal thickness, lower endothelial cell density, lower endothelial cell density, and a greater coefficient of variance, although these changes were statistically insignificant [14]. Finally, Jha et al. investigated the endothelial cell density, shape, and central corneal thickness in individuals with diabetes mellitus. They claimed that diabetes had negative effects on the thickness and endothelium of the cornea [12]. Conclusions According to recent meta-analysis findings, people with diabetes had larger average cell sizes, thicker corneas, and greater coefficients of variation. Conversely, diabetes subjects had lower endothelial cell densities and hexagonal cell ratios. Some restrictions applied to the current investigation. The search was omitted from the online databases, the non-online records were excluded, and the non-English articles were excluded. Due to a dearth of reporting, several factors, including paraclinical traits and clinical parameters, were not assessed. These may be taken into account in the upcoming network meta-analysis. Acknowledgements Not applicable. Author contributions The core idea of this study came from ASH. AMM and MN-A wrote and analyzed the collected papers. All authors wrote the manuscript. Final editing was done by MN-A. ASH and MN-A supervised the manuscript. Funding Not applicable. Data availability Not applicable. Declarations Ethics approval and consent to participate Not applicable. Consent for publication Not applicable. Competing interests There are neither ethical nor financial conflicts of interest involved in the manuscript. The manuscript is not submitted for publication elsewhere. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement Not applicable.