Tear Film Stability After Primary Pterygium Excision Combined with Limbal Stem Cell Transplantation is Superior to Amniotic Membrane Transplantation: A Meta-Analysis


 Purpose To evaluate the tear film stability after primary pterygium excision combined with Limbal stem cell transplantation (LSCT) or amniotic membrane transplantation (AMT). Methods We searched the PubMed, EMBASE, Web of Science, Cochrane Library, CNKI, Wan Fang and VIP databases for all studies on tear film stability after primary pterygium excision combined with LSCT or AMT. The mean difference (MD) and 95% confidence interval (CI) were calculated for outcomes using the fixed effect or random effect model.Results Seven studies with a total of 531 eyes were enrolled in our meta-analysis, which revealed that comparison between the LSCT group and the AMT group: Ocular Surface Disease Index (OSDI) 3 months postoperatively (MD=-5.16, 95%CI:-6.48,-3.85, P＜0.05), Tear break-up time (BUT) 1 or 3 months postoperatively (1 month: MD=0.30, 95%CI:-0.66,1.26, P=0.54; 3 months: MD=1.30, 95%CI:-0.13,2.72, P=0.07), Schirmer I test 1 or 3 months postoperatively (1 month: MD=0.05, 95%CI:-0.41,0.51, P=0.82; 3 months: MD=1.41, 95%CI:0.81,2.02, P＜0.05), Corneal fluorescein staining (CFS) score 1 month postoperatively (MD=-0.49, 95%CI:-1.29,0.31, P=0.23).Conclusion Primary pterygium excision combined with LSCT is associated with better tear film stability changes than AMT.


Introduction
Pterygium is a wing-shaped brovascular growth of the conjunctiva that extends across the limbus and invades the cornea [1]. It is a common ocular surface disease and is extremely common in occurrence and worldwide distribution [2]. The exact pathogenesis of the injury is complex and remains incompletely understood. It is associated with multiple risk factors such as ultraviolet light, age, hereditary factors, chronic in ammation, microtrauma, and heat [3].
Pterygia can affect vision and cause redness, foreign body sensation, decreased tear lm stability, and affected patients frequently have dry eye disease (DED) symptoms [4].
The only effective method for the treatment of pterygium is surgery [5]. There are many surgical methods for pterygium. Each of these methods has its pros and cons, and there is no gold standard for pterygium surgery. Common surgical methods for pterygium currently consist of surgical resection of the involved area, followed by its coverage using conjunctiva with limbal stem cells or an amniotic membrane [6]. Limbal stem cells are located in the limbal epithelial layer and are the ultimate source of the transparent corneal epithelium [7,8]. Amniotic membrane transplantation (AMT) is increasingly used in ophthalmic surgery.
It can resist adhesions and is bene cial in promoting epithelialization and inhibiting in ammation and neovascularization [9]. Another advantage of AMT application to the ocular surface is that it is an avascular tissue, which can effectively inhibit neovascularization during corneal surface reconstruction [10].
Limbal stem cell transplantation (LSCT) and AMT can both be used in pterygium surgery.
A stable anterior tear lm is essential for ocular health, mainly because it creates a protective and lubricated environment for the tissues of the palpebral bulbar surfaces [11]. Recent studies have demonstrated that pterygium has a close relationship with tear lm stability. Although decreased tear lm stability has been reported as one of the risk factors for pterygium formation, pterygium itself also contributes to ocular surface instability [12]. Pathological conjunctival, corneal or eyelid changes in pterygium can decrease tear lm stability. Patients with pterygium have disturbances in tear quality, quantity, and reduction of conjunctival goblet cell population [3]. One study showed that excision of pterygium could lead to an improved tear lm stability [13]. The main objective of this summary of our review is to report the comparison of tear lm stability after primary pterygium excision combined with LSCT or AMT based on the best available evidence.

Search Strategy
We searched the PubMed, EMBASE, Web of Science, Cochrane Library, CNKI, Wan Fang and VIP databases for all studies on tear lm stability after primary pterygium excision combined with LSCT or AMT. The Mesh and Entry terms were as follows: ("Tear lm" OR "Dry Eye Syndromes") AND ("Pterygium") AND ("Limbal stem cell") AND ("Amnion" OR "Amniotic Membrane") AND ("Transplant" OR "Graft" OR "Transplantation") AND ("randomized controlled trial" OR "randomized" OR "placebo"). The study selection process was shown in Fig. 1. The meta-analysis was performed according to the PRISMA statement and was registered at International Prospective Register of System Reviews (CRD number: 42021265281: www.crd.york.ac.uk/PROSPERO/).

Selection Criteria
The inclusion criteria were: (1) Type of studies: Randomized controlled trials (RCT); (2) Type of participants: Patients with primary pterygium;

Quality Assessment
Our team authors individually reviewed the risk of selection bias, performance bias, detection bias, attribution bias, reporting bias and other biases of each included study using the Cochrane Risk of Bias Tool. Disagreements were resolved by our team authors by discussion. The risks of bias for 7 studies were classi ed as low, high, or unclear.

Data Extraction
Extracted information and data from the selected studies were as follows: the rst author's surname, year of publication, country, study type, sample size, OSDI, BUT, Schirmer I test and CFS score. The outcome measures were continuous data, which we expressed as mean and standard deviation (SD).

Statistical Analysis
The mean difference (MD) and 95% con dence interval (CI) were calculated for outcomes. Heterogeneity was estimated using the Cochrane I square (I 2 ) statistics and Q statistics. The xed effect model was employed when low heterogeneity (I 2 < 50%, P > 0.1). The sensitivity analysis was performed when signi cant heterogeneity (I 2 > 50%, P < 0.1) was detected. The random effect model was adopted if heterogeneity could not be eliminated. Publication bias was estimated using Begg's and Egger's tests. A p-value of less than 0.05 was considered statistically signi cant. All analyses were down with RevMan version 5.3 and Stata version 15.1 software.

Search Results
The speci c selection process was shown in Fig. 1. Forty-four records were identi ed through the initial electronic search, of which 23 reports were screened after removing duplicates. Among the remaining studies, 8 were excluded because their titles and abstracts were irrelevant to our study. Then 8 studies were excluded for other reasons: incomplete data reporting or inrigorous experimental design. Finally, 7 studies [14][15][16][17][18][19][20] were enrolled in our meta-analysis.

Characteristics of Included Trials
The 7 studies nally included were published between 2013 and 2020, which provided a large sample size containing 531 eyes. Of these, 7 eligible studies were all conducted in China, which included at least one of the outcome measures: OSDI, BUT, Schirmer I test and CFS score 1 or 3 months postoperatively.
The characteristics of all included studies were shown in Table 1. Quantitative Analysis

OSDI 3 months postoperatively
Two studies were conducted in the meta-analysis of OSDI outcomes [17,18]. A total of 86 eyes were randomly allocated to the pterygium excision combined with the LSCT group and 80 to the AMT group. Compared with the AMT group, the OSDI scores were decreased in the LSCT group. The difference was statistically signi cant (pooled MD=-5.16, 95%CI:-6.48,-3.85, P 0.05, xed effect), with low heterogeneity (I 2 = 0%, P = 0.597) (Fig. 3).

BUT 1 month postoperatively
Data on BUT 1 month postoperatively were obtained from 6 studies [14-17, 19, 20], and the forest plot was shown in Fig. 4. Pooled data from 240 eyes were randomly allocated to the LSCT group and 231 to the AMT group. The heterogeneity test result showed signi cant heterogeneity across studies (I 2 = 99%, P 0.1, xed effect) (Fig. 4a). A sensitivity analysis was conducted for the 6 studies, and it was found that the study of Chen 2018 had a signi cant effect on the heterogeneity. After this study was removed, the pooled effect size of the meta-analysis varied considerably (Table 2). Therefore, the study of Chen 2018 was removed. The heterogeneity test was performed again, and the result showed that the remaining 5 studies still had high heterogeneity (I 2 = 82%, P = 0.15, xed effect) (Fig. 4b). Finally, the random effect model was adopted for the study. Since the data and results of Chen 2018 were too heterogeneous compared with the results of other studies, it was still excluded. The values of BUT were slightly increased in the LSCT group compared to the AMT group, but there was no statistical difference (pooled MD = 0.30, 95%CI:-0.66,1.26, P = 0.54, random effect) (Fig. 4c).

BUT 3 months postoperatively
Data on BUT 3 months postoperatively were obtained from 6 studies [14,[16][17][18][19][20], and the forest plot was shown in Fig. 5. Pooled data from 220 eyes were randomly allocated to the LSCT group, and 211 eyes were assigned to the AMT group. According to the heterogeneity test in this study, it was prompted that overall heterogeneity was high (I 2 = 94%, P 0.1, xed effect) (Fig. 5a). A sensitivity analysis was conducted for the 6 studies, and it was found that no matter which study was removed, the change of pooled effect size was not signi cant (Table 3). Therefore, we chose the random effect model for the study. The values of BUT were increased in the LSCT group compared to the AMT group. However, the difference was not signi cant (pooled MD = 1.30, 95%CI:-0.13,2.72, P = 0.07, random effect) (Fig. 5b).

Schirmer I test 1 month postoperatively
The data of Schirmer I test 1 month postoperatively were available from 5 studies [14][15][16][17]19], and the forest plot was shown in Fig. 6. A total of 186 eyes were randomly allocated to the LSCT group and 182 to the AMT group. The heterogeneity test was conducted, showing heterogeneity among the selected 5 studies (I 2 = 82%, P 0.1, xed effect) (Fig. 6a). A sensitivity analysis was conducted for the 5 studies, and it was found that the heterogeneity came from the study of Chen 2018 (Table 4). Therefore, the study of Chen 2018 was removed. The heterogeneity test was performed again, and the result showed low heterogeneity among the remaining 4 studies (I 2 = 50%, P = 0.11, xed effect) (Fig. 6b). Finally, the xed effect model was employed for the study. The wetting Schirmer paper strip's measured length was increased slightly in the LSCT group compared with the AMT group. However, the difference was not signi cant (pooled MD = 0.05, 95%CI:-0.41,0.51, P = 0.82, xed effect) (Fig. 6b).

Schirmer I test 3 months postoperatively
The data of Schirmer I test 3 months postoperatively were available from 5 studies [14,[16][17][18][19], and the forest plot was shown in Fig. 7. A pooled total of 166 eyes were randomly allocated to the LSCT group and 162 to the AMT group. According to the heterogeneity test, it was shown that there was signi cant heterogeneity among the 5 studies (I 2 = 82%, P 0.1, xed effect) (Fig. 7a). After conducting the sensitivity analysis, we found that when we excluded the study of Dai 2013 (Table 5), there was no heterogeneity among the remaining 4 studies (I 2 = 16%, P = 0.31, xed effect) (Fig. 7b). The increased measured length of wetting Schirmer paper strip in the LSCT group compared to the AMT group. The difference was statistically signi cant (pooled MD = 1.41, 95%CI:0.81,2.02, P 0.05, xed effect) (Fig. 7b).

CFS 1 month postoperatively
Two studies were conducted in the meta-analysis of CFS outcomes [15,16]. Seventy eyes were randomly allocated to the LSCT group and 70 to the AMT group. Compared with the AMT group, the CFS score was decreased in the LSCT group, but there was no statistical difference (pooled MD=-0.49, 95%CI:-1.29,0.31, P = 0.23, random effect) (Fig. 8).

Publication Bias
The shape of the funnel plot revealed that the data of each outcome measure were symmetrical (Fig. 9), which was con rmed by the Begg's and Egger's tests (all P 0.05). Indicating that the results of each outcome measure in our meta-analysis did not show a publication bias.

Discussion
Pterygium is a common frequently ocular surface disease with a global prevalence of 12% [21]. Besides, pterygium is a multifactorial degenerative disease [1]. However, the etiopathology remains unclear. One of the primary risk factors of pterygium is exposure to ultraviolet light. The most effective method for the treatment of this disorder is surgery. LSCT or AMT is commonly used for pterygium surgery. LSCT is better in terms of esthetic appearance and has less recurrence rate than AMT [22]. Of course, AMT also has its advantages, which can be bene cial in patients with ocular surface reconstruction, such as extensive conjunctival scarring and chemical injury or those who may require future glaucoma surgery [1,10]. A previous study showed that regardless of the surgical approach, the tear lm stability of patients after pterygium surgery is improved to varying degrees [13]. However, because the degree of tear lm stability improvement by the above two transplant means still showed con icting results [14][15][16][17][18][19][20], we conducted this meta-analysis to provide a more robust and accurate estimate of our study.
The present study elicited four main ndings through subjective tests (OSDI) and objective tests (BUT, Schirmer I test and CFS score). First, the extracted data from 7 studies revealed decreased DED symptoms (based on the OSDI) in the LSCT group compared to the AMT group (MD=-5.16, 95%CI:-6.48 -3.85, P 0.05). Second, Improved tear lm stability 1 or 3 months postoperatively (based on the BUT) in the LSCT group compared with the AMT group. However, none of the Tear lm stability is primarily assessed by BUT. Besides, OSDI, Schirmer I test and CFS score are also available for reference. As a subjective method, OSDI is widely used to evaluate DED in clinical trials, measuring the frequency of symptom appearance, environmental conditions, and quality of life-related to vision [23]. Although the score is produced subjectively, it is well-known that it has reliability and reproducibility compared with other objective tests [24]. Decreased tear lm stability can induce symptoms of DED. Therefore, subjective measures such as OSDI may be important indicators for evaluating tear lm stability.
The BUT is the most commonly used clinical diagnostic test for tear lm stability [25]. It measures the time interval between the complete blink and the appearance of the rst break in the tear lm [23]. The Schirmer I test is a commonly used method for measuring tear production [23]. Tear osmolarity is a function of tear secretion and tear evaporation [26]. Studies have shown that lower tear production can lead to higher tear osmolarity levels [27]. And there is an interconnection between hyperosmolarity and tear instability [28]. Therefore, Schirmer I test can be used as an indirect assessment indicator of tear lm stability. CFS score is an informative marker of severe DED. However, patients with mild/moderate DED showed a poor correlation with CFS score [29]. Thus, it may be poorly linked with mild tear lm instability.
This study has several limitations. First, most of the included studies had a relatively small sample size. It may be caused by the search was restricted to articles written in English or Chinese, so some compliant studies not written in English or Chinese may be excluded. Second, our meta-analysis involved only a Chinese population. One possible reason was that there were few studies in other countries on improving tear lm stability after pterygium surgery comparing LSCT with AMT.
In conclusion, our study showed that the OSDI and Schirmer I test 3 months postoperatively of the LSCT group were better than those of the AMT group with statistical signi cance. We can preliminarily suggest that tear lm stability after primary pterygium excision combined with LSCT is superior to AMT. Further studies with larger sample sizes, well-designed RCTs, and longer follow-up are needed to con rm the results reported in our meta-analysis.

Declarations
Funding This work was supported by the Key Research and Development Program of Shandong Province, China(No. 2018GSF118081).

Con ict of Interest Statement
No con icting relationship of interest exists for any author.

Data availability statement
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.       Forest plot of CFS 1 month postoperatively between the LSCT group and the AMT group