Correlation Study Of Retinal Structures And Clinical Features In Neurodegenerative Diseases

Objective:Alzheimer’s disease（AD）and Parkinson’s Disease（PD）are the most common neurodegenerative diseases in the elderly,with high morbidity,high disability rate and low diagnosis rate.Early auxiliary diagnosis of AD/PD is of great significance for delaying disease progression.The present study aimed to explore the possibility of OCT/OCTA as an auxiliary tool for diagnosis and monitoring progression in AD/PD,by evaluating changes in retinal structural characteristics and retinal microvascular density in AD/PD patients,determining their relationships with disease severity and AD/PD biomarkers,as well as developing machine learning model based on retinal structural features.Methods:（1）Clinical samples,epidemiological information,and the clinical data were collected,including age,sex,education level,disease duration,and past disease history.AD patients completed cognitive function assessment,brain magnetic resonance imaging assessment,APOE genotyping and cerebrospinal fluid（CSF）biomarker detection.Clinical assessments（including motor and non-motor symptoms）were measured across all PD patients.（2）The peripapillary retinal nerve fiber layer（p RNFL）and macular retinal thickness were measured using optical coherence tomography（OCT）.All OCT images were segmented using advanced automated 3D retinal layer segmentation software（IOWA OCTExplorer v3.8.0）,and the intraretinal layer thickness were measured.（3）And the macular vascular density and the foveal avascular zone（FAZ）area were measured by using optical coherence tomography angiography（OCTA）.（4）The correlation between the OCT/OCTA measurements and the clinical phenotype,AD/PD biomarkers were analyzed by Pearson’s correlation analysis.（5）The diagnostic performance of machine learning algorithms based on the OCT/OCTA measurements was evaluated by Python programming language.Results 1:（1）A total of 159 AD patients and 299 age-and gender-matched normal control participants completed the OCT scan of the random eyes.Among them,77 AD patients and 145 normal controls completed OCTA scans at the same time.（2）The mean,superior and inferior quadrant parapapillary nerve fiber layer（p RNFL）thickness,and macular nerve fiber layer（m RNFL）,ganglion cell layer（GCL）,inner plexiform layer（IPL）thicknesses were all significantly decreased in the AD group compared to the healthy controls group.In addition,AD group had significantly lower macular retinal thickness（MRT）and lower total macular volume（TMV）compared to the healthy controls group.Compared with the normal control group,the GCL and IPL thickness were significantly thinner in the mild AD group.In addition,the mean,superior p RNFL,m RNFL,GCL,IPL thickness,and MRT and TMV values began to demonstrate significantly decreased in the moderate AD group,and the inferior,nasal and temporal p RNFL thickness was found to be significantly decreased only in the severe AD group（all p-values less than 0.05）.（3）The macular vascular density（m-VD）of the AD group was significantly declined than that of the healthy controls group in all nine Early Treatment Diabetic Retinopathy Study（ETDRS）subfields（all p<0.05）.Compared with the healthy controls group,the mild AD group showed lower m-VD in the fovea,superior inner,inferior inner,nasal inner subfields and the whole inner ring（p<0.05）.Moreover,the m-VD exhibits a broader abnormality with increasing severity of AD.And,in the severe AD group,we found that m-VD in all ETDRS regions and the whole inner ring,the whole outer ring was significantly lower than those in the healthy controls group（p<0.05）.However,no significant difference in the FAZ area was noted regardless of comparison between AD and healthy controls,or comparison among AD subgroups（p>0.05）.（4）The superior quadrant p RNFL thickness,MRT and TMV were positively associated with Mini-Mental state Examination（MMSE）scores.And the mean,superior and inferior quadrant p RNFL,MRT and TMV were positively associated with Montreal Cognitive Assessment（Mo CA）scores.In addition,we observed a significant inverse association of the medial temporal lobe atrophy（MTA）score with the superior quadrant p RNFL thickness,MRT,IPL thickness and TMV in AD patients.Moreover,the GCL thickness correlated negatively with the parietal cortical atrophy（PCA）score,and MRT and TMV were correlated negatively with Fazekas score.However,no significant correlations were observed between any OCT measure and the global cortical atrophy（GCA）score.Correlation analyses between OCT measurements and AD biomarker results showed that the GCL thickness was positively correlated with cerebrospinal fluid Aβ₄₂/Aβ₄₀and negatively associated with CSF p-tau concentrations in the AD group.The IPL thickness was positively correlated with CSF Aβ₄₂concentrations and negatively associated with CSF t-tau concentrations in the AD group.However,no association was found between CSF biomarkers and p RNFL thickness.All retinal parameters were decreased in the APOEε4-carrier group compared to the APOEε4-noncarrier group,though this difference was only significant in IPL thickness（p=0.012）.（5）The Pearson’s correlation analysis showed that m-VD in superior inner,inferior inner,temporal inner,inferior outer,and temporal outer subfields were positively associated with MMSE scores in AD participants.The m-VD in superior inner was positively associated with MOCA scores.In addition,the m-VD of the whole inner ring was positively associated with MMSE and Mo CA scores.We observed significant inverse associations between m-VD in superior outer,inferior outer subfield and MTA score.Moreover,the m-VD in the fovea,superior inner,nasal inner subfield,and m-VD of the whole outer ring were correlated negatively with the Fazekas score.No significant correlations were observed between m-VD and the PCA,GCA score,and between m-VD and CSF biomarkers.The m-VD in the fovea,superior inner,inferior inner,temporal inner,inferior outer,and temporal outer subfields and m-VD of the whole inner ring were decreased in the APOEε4-carrier group compared to the APOEε4-noncarrier group.（6）Among the six machine learning models constructed based on OCT parameters of AD patients,the extreme gradient boosting（XGBoost）machine learning algorithm showed the best AD diagnosis performance.Among the six machine learning models constructed based on OCTA parameters of AD patients,the random forest algorithm showed the best AD diagnostic performance.Results 2:（1）In addition,397 PD patients and 427 age-and sex-matched normal controls completed OCT scan.Among them,115 PD patients and67 normal controls completed OCTA scans at the same time.（2）The mean and temporal quadrant p RNFL thicknesses,TMV,MRT,m RNFL,GCL,IPL,inner nuclear layer（INL）,and outer nuclear layer（ONL）thicknesses were significantly lower in the PD group than in the healthy controls group（p<0.05）,while the outer plexiform layer（OPL）layer thickness was significantly thicker compared with the normal control group（p<0.05）.Furthermore,significant thinning of the GCL,IPL,and ONL thickness could be detected early in patients with PD with the Hoehn-Yahr（H-Y）1 stage.In addition to the above intraretinal layers,we found that significant mean、temporal quadrant p RNFL thickness、MRT、m RNFL、INL thickness and TMV decreasing,and OPL thickening occurred in patients with PD with H-Y 2 stage.While significant thinning in superior and inferior quadrant p RNFL thickness were only detectable in PD patients with H-Y 3 stage.（3）Compared with healthy controls groups,m-VD in PD group declined noticeably in all subfields（all p<0.05）.Moreover,statistical analyses indicated the FAZ area decreasing significantly in the superficial capillary plexus of the PD patients in comparison with the healthy controls group.In addition,results suggested that PD patients with H-Y 1stage showed the vascular density of the fovea and FAZ area noticeably declining early（p<0.05）.Furthermore,results showed the decrease in m-VD with the increase in H-Y stage.The vascular density in all grids in PD patients with H-Y 3 was significantly lower than those in healthy eyes（p<0.05）.（4）Pearson’s correlation analysis revealed a negative correlation between average,temporal quadrant p RNFL thickness and Movement Disorder Society-Unified Parkinson’s Disease Rating Scale III（MDS-UPDRS III）scores（p<0.05）.In addition,the MRT,m RNFL,GCL,IPL,INL,ONL thickness,and TMV were inversely correlated with MDS-UPDRS III scores（all p<0.05）.The average p RNFL,m RNFL and GCL thickness were positively associated with Mini-Mental State Examination（MMSE）scores.There was an inverse correlation between the average p RNFL,m RNFL,GCL,IPL,INL thickness,MRT,TMV,and rapid eye movement sleep behavior disorder questionnaire–Hong Kong（RBDQ-HK）scores.The m RNFL,GCL,IPL thickness,and MRT and TMV were inversely correlated with Epworth sleepiness scale（ESS）scores.In addition,there was a positive association between GCL,IPL thickness,TMV,and HRS scores.The GCL and INL thickness,and MRT and TMV were negatively correlated with Hamilton rating scale for depression（HAMD）scores.The MRT,m RNFL,GCL,IPL,INL,and TMV of PD combined with constipation group were significantly lower than those of PD without constipation group.（5）In addition,there was a negative correlation was identified between the m-VD in superior inner,temporal outer subfields and MDS-UPDRS III scores.The m-VD in the fovea,superior inner,inferior inner,nasal inner,temporal inner,nasal outer,temporal outer subfields,and FAZ area were significantly negatively correlated with H-Y staging scores.In addition,the m-VD in the superior inner,inferior inner,and temporal inner subfields were significantly positively correlated with the MMSE score.There was an inverse correlation between the m-VD in nasal inner and RBDQ-HK scores.The m-VD in the fovea,temporal inner,temporal outer subfields were inversely correlated with ESS scores.In addition,there was a positive association between FAZ area and HRS scores.The m-VD in superior inner,inferior inner,nasal inner,temporal inner,superior outer,inferior outer,nasal outer and temporal outer subfields of PD combined with constipation group were significantly lower than those of PD without constipation group.（6）Among the six machine learning models constructed based on OCT and OCTA parameters of PD,the XGBoost algorithm was superior to the traditional Logistic Regression and other five machine learning algorithms and exhibited the best PD diagnostic performance.Conclusions:（1）The retinal thickness and macular microvascular density were decreased in AD/PD patients.（2）The retinal thickness and macular microvascular density in AD patients were correlated with MMSE,Mo CA scores,brain MRI visual rating scale,cerebrospinal fluid biomarkers(Aβ₄₂、p-tau、t-tau、Aβ₄₂/Aβ₄₀),and APOE genotypes.（3）The retinal thickness and macular microvascular density in PD patients were significantly correlated with H-Y stage,MDS-UPDRS III,RBDQ-HK,ESS,HRS,HAMD scores and and constipation symptoms.（4）OCT,OCTA could become an auxiliary tool for diagnosis and monitoring progression in AD/PD.