Effects And Mechanisms Of Diabetes Mellitus And Panretinal Photocoagulation On The Corneal Subbasal Nerve Plexus

ObjectiveDiabetic corneal nerve damage,which is considered to be a part of the diabetic peripheral neuropathy(DPN),can occur in the early stage of diabetes mellitus(DM).However,limited by the small investigated area and poor topographic reproducibility of the examination equipment-corneal confocal microscopy(CCM),its evaluation often has large bias,which leads to difficulty in establishing a threshold for diagnose;on the other hand,it has not yet reached a consensus that whether the panretinal photocoagulation(PRP)treatment for diabetes retinopathy(DR)-the most serious DM complication in the eye,will cause further damage to the corneal nerve.Therefore,there is an urgent need to improve the CCM technology in order to make up for limitations of the equipment,and improve the accuracy and reliability of the examination,so then break through the bottleneck of current research.The purpose of the current study is to improve the CCM technology and synthesize the standard single image into corneal subbasal nerve plexus(SNP)mosaics.On this basis,quantitative analysis of nerve parameters was carried out to evaluate the effect of DM and PRP treatment on SNP,and to explore the possible pathological mechanisms,so as to provide theoretical basis for clinical prevention and treatment of related ocular surface diseases.Methods1.Mosaicking of the normal human corneal SNP:30 healthy volunteers between the age of 20 and 30 years old were recruited for a cross-sectional observational study.They were all cleared of systemic and local diseases which may affect the nerve system.Their right eyes were selected as the test eyes,and were scanned with the Heidelberg Retinal Tomograph Ⅱ(HRT-Ⅱ)to collect images of the SNP over an area of 2～3 mm around the whorl-like complex.Captured images were arranged into a contiguous montage by using Adobe Photoshop.Two experienced ophthalmologists selected three representative,clear,complete and no-overlapping standard single images in the whorl-like region respectively;and then selected a standard frame size of 700 μ m × 700 μ m with whorl-like complex as the center in the montage respectively.The selected images were then analyzed by semi-automatic image analysis system Image J.Coefficient of repeatability(CoR)and Bland-Altman analysis were used to evaluate the inter-observer agreement of the traditional multi-standard image approach and the mosaic approach respectively,so as to verify the application value of the mosaic approach for quantitative analysis of the SNP.2.Characteristics of corneal SNP in DR:Patients who were aged between 50 to 65 years old with type 2 diabetes were included from Shanxi Eye Hospital,and were divided into non-diabetic retinopathy group(NDR),non-proliferative diabetic retinopathy group(NPDR)and proliferative diabetic retinopathy group(PDR)according to their conditions.Age and gender matched age-related cataract patients were also included as control group.All the subjects underwent corneal confocal microscopy(CCM)to observe the morphology of the corneal subbasal nerve plexus(SNP)in the whorl-like region,and measure the nerve fiber length(NFL)in this area.Covariance analysis and Spearman rank correlation analysis was used to estimate the correlation between DR stages and diabetic corneal nerve damage.3.Changes and mechanisms of corneal SNP in diabetic patients subjected to PRP:Type 2 diabetic patients who were aged 50 to 65 years old with binocular diabetic retinopathy stage Ⅳ and waited for PRP treatment were included.They were divided into horizontal laser group and vertical laser group according to the random number method.Their severe eyes were chosen as the treatment eye and the contralateral eyes were chosen as the control eye.CCM was performed before PRP treatment,1 week after every photocoagulation for 4 times,and 1 month after the completion of PRP to examine the morphology of SNP and Langerhans cell(LC)over an area of 2～3 mm around the whorl-like complex,and measure the of NFL and LC density.OCT was performed to measure the thickness of choroid in the macular area before and 1 month after the the PRP treatment.Repeated measurement variance analysis,MIXED model of SAS software and Pearson correlation analysis were used to explore the effects of PRP on the SNP and its related mechanisms.Results1.We successfully performed wide-field SNP mosaics of thirty healthy volunteers according to the independently designed examination standard combined with a method described in a previously published study.The whorl-like complex was observed inferior to the central cornea in all the examined eyes.Images selected by the mosaic approach showed fairly uniform in nerve morphology,with excellent inter-observer agreement indicated by CoR=3.79%,the clustering of NFL differences around the value of 0,and a narrow 95%limit of agreement(LOA)in Bland-Altman scatter plots;while images selected by the traditional multiple standard image approach showed great variation,and poor inter-observer agreement with CoR=11.64%,the dispersed distribution of NFL differences,and a wider 95%LOA.2.Comparing with the normal control group,nerve fibers in whorl-like region in all the DR groups showed generally decrease in diameter and density,with increase in tortuous,accompanied by different degrees of whorl-like structure loss.The NDR group and NPDR group mainly showed loss in the whorl-like center.Some eyes in the PDR group also showed loss of inferior and temporal structures,in addition to the loss in the whorl-like center.The NFL was 21.08±4.74mm/mm2 in the control group,16.47±6.35mm/mm2 in the NDR group,14.95±3.90mm/mm2 in the NPDR group,and 11.61±3.24mm/mm2 in the PDR group.The difference between groups was statistically significant(F=10.54,P<0.001).The NFL comparison between NDR group and PDR group was statistically significant(P=0.008),while the comparisons between NDR group and NPDR group,and between NPDR group and PDR group were not statistically significant(P=0.359 and P=0.075,respectively).Spearman rank correlation analysis showed that there was a significant negative correlation between NFL and DR stages(r=-0.582).3.After PRP treatment,SNP nerve fibers showed decrease in diameter and density,accompanied by different degrees of neural structure loss in some patients.The overall comparison of NFL values between observation time points of the treated eyes was statistically significant(F=4.241,P=0.010).In the horizontal laser group,SNP nerve fiber density decreased after the first and second photocoagulation,and The overall comparison of NFL values between observation time point was statistically significant(F=7.192,P<0.001),among which the differences between before PRP treatment and 1 week after the first photocoagulation,1 week after the second photocoagulation,and 1 month after the completion of PRP were statistically significant(all P<0.05);In the vertical laser group,the SNP nerve fiber density decreased after the third and fourth photocoagulation,but the overall comparison of NFL values between observation time points was not statistically significant(F=1.436,P=0.245).After PRP treatment,LC was increased,clustering around the whorl-like are,and the absence of SNP neural structure was seen in the mature LC infiltration area.The overall difference in LC density between observation time points was statistically significant(F=14.640,P<0.001),among which the differences between before PRP treatment and 1 week after the second photocoagulation,1 week after the third photocoagulation,1 week after the fourth photocoagulation and 1 month after PRP was statistically significant(all P<0.05).There was no significant difference in the overall LC density at each observation time point between the horizontal laser group and the vertical laser group(F group×time=0.001,P group×time=0.971).The overall comparison of immature LC density between observation time points was statistically significant(F=17.800,P 0.001),while the overall comparison of mature LC density between observation time points was not statistically significant(F=1.200,P=0.290).Correlation analysis showed that the overall LC density and immature LC density after the completion of PRP were significantly positively correlated with their corresponding baseline levels(r overall LC=0.674,P<0.001;r immature LC=0.712,P<0.001),but the mature LC density after the completion of PRP had no significant correlation with its baseline level(r mature LC=0.187,P=0.152).There was a negative correlation between NFL value and LC density(ρ=-0.041).But there was no significant correlation neither between NFL difference before and after PRP treatment and baseline choroidal thickness,nor between NFL difference and choroidal thickness difference before and after PRP treatment(r baseline=-0.009,P=0.946;r difference=0.007,P=0.960).Conclusions1.According to the independently specific examination standard,wide-field SNP mosaics can be successfully performed using the existing CCM equipment,without adding any additional hardware or software,under the premise of solving several technical problems such as how to fix a site for landmark,how to complete continuous scan and how to control the image quality.2.The mosaic approach can provide a significantly larger area of view,which enable the researchers to obtain more morphological information of SNP,and makes it possible to repeat scanning of the same site.Its agreement for neural parameter assessment is also better than the traditional multi-standard image method.3.Diabetic patients showed SNP decrease in nerve diameter and density,with increase in tortuous,accompanied by different degrees of the nerve structure loss,and the loss begins from the whorl-like center and gradually develops downward and the temporal side.4.Diabetic nerve damage in SNP occurs earlier than DR,and progressed gradually in the development of DR,there was a negative correlation between NFL and the severity of DR.5.PRP treatment can lead to SNP damage in DR patients,which is characterized by decrease in nerve fiber,diameter and density,and loss of nerve structure in different degrees;6.SNP was prone to been damaged when performing horizontal photocoagulation,which supports the inference that PRP affects the corneal SNP function by damaging the posterior long ciliary nerve through heat conduction;7.LC density was increased,and the absence of SNP neural structure was seen in the mature LC infiltration area,which supports the theory that immune mechanism participates in the corneal nerve damage subjected to PRP treatment.