Meta-analysis of corneal endothelial changes after phacoemulsification in diabetic and non-diabetic patients

The study included prospective studies. We included patients (1) with and without diabetes who underwent phacoemulsification and intraocular lens implantation, (2) whose outcomes included at least one data index of corneal properties (CCT, ECD, CV, and HCP), (3) with no other systemic diseases except DM, (4) whose blood glucose levels were stable, and (5) with no serious surgery-related complications. Patients with severe ocular and systemic complications caused by DM were excluded, such as those with proliferative diabetic retinopathy (PDR) and diabetic nephropathy. Those with mature cataracts (brown/white), cataract grade V, or other eye diseases were also excluded.

We selected relevant studies published between January 1, 2011 and December 25, 2021, by searching the databases PubMed, Embase, Web of Science, and the Cochrane Library (Trials Central). No language restrictions were applied. We used the following MeSH terms and Text Words: The complete search used for PubMed was: (((“Cataract”[Mesh]) OR (Cataracts [Title/Abstract])) OR (Lens Opacity* [Title/Abstract])) OR (Opaciti*, Lens [Title/ Abstract])) OR (Cataract*, Membranous [Title/Abstract])) OR (Membranous Cataract* [Title/Abstract])) OR (Pseudoaphakia [Title/Abstract])) OR (Phacoemulsification* [Title/Abstract]))) AND ((“Diabetes Mellitus”[Mesh]) OR (diabete* [Title /Abstract]) OR (diabetic* [Title/ Abstract])) AND ((“Cornea”[Mesh]) OR (Cornea* [Title/ Abstract])). Filters: from 2011/1/1 to 2021/12/25. Manual search was conducted on the reference lists of published key articles in English.

The quality of the selected studies was assessed using the Newcastle–Ottawa Scale (NOS) CASE CONTROL STUDIES, which includes three sections: selection (four items, four points), comparability (one item, two points), and exposure (three items, three points); a total of nine points is achievable, with scores ≥ 6 indicating good quality. Detailed items for the NOS are provided in Additional file 1.

Two independent investigators extracted the following information: first author and country, publication year, type of study, follow-up duration, patient age, number of eyes, ascertainment criteria for DM and cataracts, DM status (duration or fasting blood sugar or glycated hemoglobin [HbA1c]), presence of diabetic retinopathy, and literature quality assessment scores.

A forest plot was constructed and statistical and sensitivity analyses were performed using Review Manager 5.4.1. Sensitivity analysis was performed using the one-by-one exclusion method. The weighted mean difference (WMD) and 95% confidence interval (CI) were calculated based on selected outcomes. P < 0.05 was considered a statistically significant difference. I² test and Cochran’s Q test were used to evaluate heterogeneity. No heterogeneity was indicated by I² < 50% and P > 0.1, and the fixed-effect model was used to calculate pooled effect. If there was significant heterogeneity, a random effect model was used.

The literature selection process is shown in Fig. 1. In total, 1042 relevant studies (PubMed 132, EMbase 417, Web of Science 489, and Cochrane Library 4) were retrieved. Next, they were screened based on redundancy (801 studies remained), screening of topics (43 studies remained), and abstract information (22 studies remained). Nine studies were excluded after reading the full text: one with unknown glycemic control, one with an incomplete outcome index, two in which patients had serious DM complications (PDR surgery history, kidney disease dialysis history), two in which the basic information was not comprehensive, and three retrospective studies. Finally, 13 studies [13‐25], including 1744 eyes (788 eyes in the DM group and 956 eyes in the non-DM group), were selected for this meta-analysis.

According to the NOS, eight studies scored 7, and five studies scored 8. All studies scored more than 6 points, indicating that the quality of the included studies was high. The characteristics of the included studies are provided in Table 1.

Sensitivity analysis publication bias analysis showed that the data of Li [13] were extremely unstable regarding the CV 3 months postoperatively and HCP preoperatively and 1 month and 3 months postoperatively; thus, these data were excluded from this analysis. Although partial results showed relatively large heterogeneity, the data were stable and reliable after sensitivity analysis. Sensitivity analysis was not performed at 6 months postoperatively owing to the small number of included studies.

The Egger’s test was used to estimate publication bias. Because of the low number of included studies for CV and HCP at 6 months postoperatively, no publication bias analysis was performed at this stage. There was no publication bias in the included studies, except for preoperative CCT (Table 2).

The influence of preoperative CCT publication bias on the interpretation of the results was evaluated using the trim-and-fill method. The pooled effect sizes calculated by the fixed-effect model (pooled effect size: standard error of effect size) were 0.041 and 0.009, and the 95% CI was -0.071 to 0.152 and -0.099 to 0.116 before and after using the trim-and-fill method, respectively. No significant difference was found before and after using the trim-and-fill method (P = 0.478, P = 0.874, respectively). There was no asymmetry in the funnel plot after supplementing two studies (Supplemental studies are shown as “square” in Fig. 6). This showed that publication bias had little effect on the results, and the results were relatively stable (Fig. 6).

The hydration balance is regulated by the CEC pump in normal conditions. When the CEC pump is dysfunctional, the corneal stroma accumulates water, and swelling occurs, which is manifested by an increase in corneal thickness. However, persistent corneal edema and dysfunction do not occur unless the CEC number declines to < 500–1000 cells/mm [35].

The CCT of patients with DM has been found to be significantly higher than that of healthy individuals, and HbA1c is found to be positively correlated with CCT and CV and negatively correlated with ECD in patients with DM [35, 36]. The duration of DM has a significant impact on these parameters: the longer the DM duration, the higher the CCT and the lower the ECD [37]. In the present study, we found that CCT in the DM group was significantly higher than that in the non-DM group at the early postoperative period (3 months), suggesting that the impairment degree of corneal endothelial barrier function in the DM group was significantly higher than that in the non-DM group. From 3 to 6 months postoperatively, the difference in CCT between the two groups gradually decreased, indicating gradual recovery of the corneal endothelial function. Thus, the corneal endothelial barrier function was impaired at the early postoperative period and then gradually stabilized until 6 months postoperatively. This may be related to postoperative oxidative stress and inflammation response. DM itself [38] and surgical trauma [39] increase the oxidative stress level of CECs. Oxidative stress not only directly damages CECs [29] but also induces inflammation through multiple activation pathways [40]. Corneal edema alleviates with a decrease in inflammation, resulting in a lower CCT during the recovery process after phacoemulsification [15].

The percentage of endothelial cell loss (ECL%) in patients with DM was reported to be significantly higher than that in the control group after phacoemulsification [41, 42], and the damage was not restored to the preoperative state at 6 months postoperatively [14]. Joo et al. [43] found that the ECL in patients with DM was higher than that in non-DM patients 1 year after phacoemulsification, although not statistically significant. Furthermore, the duration of DM may affect postoperative ECD loss, with more ECD loss occurring when the duration is ≥ 10 years. Choi et al. [34] found that ECL% was about 2.06 ± 1.36% per year 10 years after phacoemulsification, and this persistent ECL may be related to corneal endothelial remodeling. Ganesan et al. [15] considered inflammation to be a risk factor for ECL in DM patients, whereas age and effective phacoemulsification time were the risk factors in non-DM patients after phacoemulsification.

Our results showed that there was no significant difference in ECD between the DM group and non-DM group preoperatively. However, the ECD in the DM group was significantly lower than that in the non-DM group and the ECL increased progressively compared with that in the non-DM group at 1–6 months postoperatively. This indicated that ECL was accelerated, which was unstable at 6 months postoperatively, and postoperative corneal recovery was delayed in patients with DM. Although the ECL% in patients with DM was higher than that in the control group after phacoemulsification, the intraoperative cumulative dissipated energy (CDE), fluid consumption, and operative time were not statistically significant [41]. The higher ECL postoperatively may be related to the advanced age of patients, increased cataract density, increased endothelial cell vulnerability in diabetic patients, increased trauma during cataract surgery, and grade of cataract [14, 16, 41].

CV and HCP reflect the dynamic repair and healing process of CEC morphology after injury; the increase in CV indicates a large variability in cell size, and the decrease in HCP indicates an increase in pleomorphism. The remaining cells expand and slide after endothelial cell injury, which shows an increase in CV and a decrease in HCP. The morphology of CECs in patients with DM was unstable at 4 weeks after phacoemulsification [17]. The HCP of patients with DM decreased significantly 3 months postoperatively, whereas the CV showed no significant difference [18]. However, some studies reported that the CV of patients with DM was significantly higher than that of those without DM at 3 months postoperatively, although this difference did not affect the corneal function [19]. The HCP returned to its preoperative state 6 months postoperatively [14]. No significant change was found in the CV and HCP in either group at 1 year postoperatively [43].

Our results showed that the degree of morphological variation of CECs in diabetic patients was largest at 1 month postoperatively, which was significantly higher than that in non-diabetic patients, and subsequently, the degree of morphological variation of CECs gradually decreased. The corneal morphology of diabetic patients was more unstable in the early postoperative stage, indicating that the endothelial cells of diabetic patients have a weaker repair ability upon damage, and the repair process takes longer [20].

Despite the fact that there was no significant difference in the visual acuity between DM and non-DM patients after phacoemulsification was performed in controlled blood glucose levels [18], the impact of diabetes on corneal health cannot be ignored, as good control of blood glucose is frequently lost in DM. Compared with healthy individuals, the CECs of patients with DM have a lower tolerance to phacoemulsification, are more likely to be damaged, and take longer time to recover, which requires the surgeon to carefully protect the cornea in order to minimize corneal endothelial damage intraoperatively. Femtosecond laser-assisted cataract surgery (FLACS) is reported to cause less damage to the corneal endothelium in patients with DM and can reduce the ECL. This may be because the corneal endothelial injury caused by the small energy during FLACS is insufficient to cause significant damage [44]. Therefore, FLACS may be a better option for patients with DM than conventional phacoemulsification. Furthermore, it should not be ignored that age and DM status are important factors affecting corneal ECD. For DM patients who require cataract surgery, timely surgery is also important when blood glucose is well controlled.

Our study has some limitations. First, intraoperative CDE, fluid consumption, and operative time were not assessed, although most of the included studies showed no statistical differences between the two groups. Second, a randomized controlled trial could not be performed because of DM presence. Third, the cataract grade and expertise of operating surgeon cannot be standardized across all studies, which may be a confounding factor in this study. Finally, longer studies after 6 months, as well as subgroup analyses of DM status (such as DM duration and HbA1c level), were not performed because few studies were eligible for inclusion. In the future, we will continue to focus on the long-term dynamic changes in corneal properties after cataract surgery in patients with DM.