The relationship between retinal layers and brain areas in asymptomatic first-degree relatives of sporadic forms of Alzheimer’s disease: an exploratory analysis

Two main genetic risks for sporadic Alzheimer’s disease (AD) are a family history and ɛ4 allele of apolipoprotein E. The brain and retina are part of the central nervous system and share pathophysiological mechanisms in AD. We performed a cross-sectional study with 30 participants without a family history of sporadic AD (FH−) and noncarriers of ApoE ɛ4 (ApoE ɛ4−) as a control group and 34 participants with a family history of sporadic AD (FH+) and carriers of at least one ɛ4 allele (ApoE ɛ4+). We analyzed the correlations between macular volumes of retinal layers and thickness of the peripapillary retinal nerve fiber layer (pRNFL) measured by optical coherence tomography (OCT) with the brain area parameters measured by magnetic resonance imaging (MRI) in participants at high genetic risk of developing AD (FH+ ApoE ɛ4+). We observed a significant volume reduction in the FH+ ApoE ɛ4+ group compared with the control group in some macular areas of (i) macular RNFL (mRNFL), (ii) inner plexiform layer (IPL), (iii) inner nuclear layer (INL), and (iv) outer plexiform layer (OPL). Furthermore, in the FH+ ApoE ɛ4+ group, the retinal sectors that showed statistically significant volume decrease correlated with brain areas that are affected in the early stages of AD. In the same group, the peripapillary retinal nerve fiber layer (pRNFL) did not show statistically significant changes in thickness compared with the control group. However, correlations of these sectors with the brain areas involved in this disease were also found. In cognitively healthy participants at high genetic risk of developing sporadic forms of AD, there are significant correlations between retinal changes and brain areas closely related to AD such as the entorhinal cortex, the lingual gyrus, and the hippocampus.


Background
Alzheimer's disease (AD) is a neurodegenerative disease and the most common cause of dementia [1]. The progressive accumulation of amyloid-beta (Aβ) protein (plaques) outside the neurons [2] and the accumulation of tau protein (tangles) within the neurons [1] are the hallmarks of this disease. Aβ aggregation starts in the anterior cingulate cortex and the precuneus [3,4], while tau tangles are located in the entorhinal cortex, the hippocampus, and adjacent limbic structures in milder cases [5]. The use of biomarkers such as quantification of Aβ and tau levels in cerebrospinal fluid (CSF) or the increased deposition of tau tangles and accumulation of Aβ plaques revealed by positron emission tomography (PET) are biomarkers that help determine the clinical stage of AD; however, they are limited because they require invasive techniques and expensive diagnostic tools [6].
The retina and the brain are part of the central nervous system, and both have a common embryological origin [7]. Currently, it is known that there is a relationship between age-related retinal neurodegenerative diseases and brain neurodegenerative diseases, including AD [8]. In addition, in the retina, protein deposits have been detected in both AD animal models and in vivo and postmortem eyes from human AD patients [9][10][11], with the retina having important diagnostic implications in this disease. Through optical coherence tomography (OCT), a reliable non-invasive diagnostic tool that is commonly used in ophthalmology to visualize and analyze the retinal layers, retinal changes have been observed in different stages of AD. In a previous work using OCT, we have demonstrated that, in preclinical stages, participants with a high genetic risk of developing AD show a significant thinning in several retinal layers in the macular region [12]. In patients with mild AD, thinning also occurs principally in the macular region [13][14][15]; when the disease progresses to a moderate stage, these changes are reflected in the peripapillary region [16].
A direct correlation has been observed between the volumes of brain areas measured by magnetic resonance imaging (MRI) and the thickness of specific retina regions using OCT in nondemented older adults [6,17]. In addition, both family history of the disease (FH+) and ApoE ɛ4 genotype (ApoE ɛ4+) potentiate each other, contributing to the thinning of the cortex in the hippocampal region [18]. The purpose of our study was to analyze the correlations between the macular volumes of all the retinal layers and the thickness of the peripapillary retinal nerve fiber layer (pRNFL) measured by OCT, with the volumes and thickness of different brain areas measured by MRI in participants at high genetic risk of developing AD (FH+ ApoE ɛ4+) compared with a control group (FH− ApoE ɛ4−).

Participants
The participants were recruited from the study "The cognitive and neurophysiological characteristics of participants at high genetic risk of developing dementia: a multidimensional approach" (COGDEM Study, PSI2015-68793-C3-2-R, Ministry of Economy and Competitiveness) (n = 251). In this study, participants were recruited in different local hospitals and advertisements were published in different professional associations such as the "Asociación Española de Ingenieros de Telecomunicación Delegación de Madrid. " The general inclusion criteria are listed below: (i) age between 45 and 75 years; (ii) absence of psychiatric or neurological disorders; (iii) absence of evidence of infection, infarction, or focal lesions on T2-weighted magnetic resonance imaging (MRI); (iv) absence of substance addictions such as alcohol or chronic use of anxiolytic, neuroleptic, narcotic, anticonvulsant, or sedative-hypnotic drugs; and (v) absence of memory complaints. In addition, all participants underwent a neuropsychological assessment including analysis of declarative memory with the Logical Memory (immediate and delayed) and digits (forward and backward) and the Word List of the Wechsler Memory Scale-III-and everyday memory with the Rivermead Behavioural Memory Test. Subjective cognitive impairment was assessed with the MFE and a series of questions (neuropsychological interview) on possible impairment of different cognitive functions We analyzed two groups. As a control group (n = 30), we selected from the COGDEM study database those participants with no family history of sporadic senile onset AD. These participants had the following characteristics: (i) they were cognitively healthy, (ii) they had no first-degree family history of AD (FH−), (iii) they were noncarriers of the ApoE ɛ4 allele (ApoE ɛ4−), and (iv) they had no memory complaints. The group at high genetic risk of developing AD (n = 34), matched with the control group in terms of age and socioeconomic status, consisted of participants with the following features: (i) cognitively healthy, (ii) firstdegree family history of senile AD (FH+) (father or mother affected by the disease), (iii) carrier of at least one ɛ4 allele for the ApoE gene (ApoE ɛ4+), and (iv) no memory complaints. Figure 1 shows a flow diagram of the patient selection.
In both groups, Mini-Mental State Examination (MMSE) scores were normal (above 26) and participants had no previous history of neurological or psychiatric disorder.
The research followed the tenets of the Declaration of Helsinki and the study protocol was approved by the Local Ethics Committee (Hospital Clinic San Carlos) with the internal code 18/422-E_BS. All participants signed written informed consent to be enrolled in the study.

Optical coherence tomography (OCT)
After an ophthalmological examination performed at the Clinic of the Ramon Castroviejo Institute of Ophthalmic Research in Madrid, which included visual acuity, slit-lamp examination, applanation tonometry (Perkins MKII tonometer, Clement Clarke International, Essex, UK), and a dilated funduscopy, one eye of each patient was randomly selected. All the participants met the following ophthalmological inclusion criteria: (i) free from ocular disease or posterior pole pathology (macular degeneration, drusen, glaucoma, or suspicion, epiretinal membrane, and congenital malformation), (ii) having a best corrected visual acuity more than 20/40, (iii) having less than ± 5 D spherocylindrical refractive error, and (iv) having an intraocular pressure less than 20 mmHg.
To analyze the volumes of the macular region, OCT imaging was performed with Spectralis OCT (Heidelberg Engineering, Heidelberg, Germany), following the OCT protocol previously published [12].
Heidelberg segmentation software (Heidelberg, Germany, version 1.10.4.0) allowed us to measure the total retinal volume and the volume of each retinal layer in the macular area. To avoid segmentation problems, all The cognitive and neurophysiological characteristic of participants at high risk of developing dementia: a multidimensional approach"; FH−, participants without a family history of Alzheimer's disease; FH+, participants with a family history of AD; ApoE ɛ4-, noncarriers of ɛ4 allele; ApoE ɛ4+, carriers of at least one ɛ4 allele. In gray are the participants who participated in the ophthalmological study and were included in this study. OHT, ocular hypertension; OCT, optical coherence tomography automatic segmentations were checked by the same optometrist (IL-C) and modified manually if an error was found. The macular volume of the following retinal layers was analyzed: retinal nerve fiber layer (RNFL), ganglion cell layer (GCL), inner plexiform layer (IPL), inner nuclear layer (INL), outer plexiform layer (OPL), outer nuclear layer (ONL), and retinal pigment epithelium (RPE). The macular area was analyzed according to the standard Early Treatment Diabetic Retinopathy Study (ETDRS) macular grid (a foveal area of 1 mm in diameter, 1-3 mm around the fovea in the inner ring, and 3-6 mm for the outer ring) [19] ( Fig. 2A). A peripapillary RNFL (pRNFL) thickness analysis was carried out in six sectors (temporal (T), superotemporal (ST), inferotemporal (IT), nasal (N), inferonasal, and superonasal (SN) (Fig. 2B). An average of all sectors was also obtained (global). In this study, we included macular volume scans with a minimum signalto-noise ratio of 25 and an average of 16 B-scans. The pRNFL scans have a minimum signal-to-noise ratio of 20 and an average of at least 40 B-scans. According to the calibration provided by the manufacturers, the thickness measurements were provided in microns and the volume measurements in mm 3 .

Magnetic resonance imaging (MRI)
For each subject, MRI were acquired using a 1. flip angle 12°; 1 mm slice thickness, 256×256 matrix and FOV 25 cm). In addition to whole-head 3D fast spoiled ir-prepped gradient-echo T1-weighted 1-mm 3 isotropic sequence, 3D CUBE FLAIR T2-weighted 1.6-mm-thickness sequence and 2D gradient-echo T2 sequence, DTI images were also obtained. T2 sequences do not contribute to the study except to rule out vascular lesions and vascular load or other differential diagnostic entities, apart from being necessary for the overall volumetric assessment (Fig. 3A, B). Each T1-weighted MRI image was processed using FreeSurfer 6.0 recon-all procedure [20][21][22][23][24]. This procedure performs a motion correction, corrects for intensity nonuniformity, and performs an intensity normalization. Then, it performs a segmentation of the different brain tissues and it constructs a cortical surface mesh for each T1. It registers an inflated sphere version of this cortical mesh to a common surface-space and it uses the Desikan-Killiany anatomical atlas [25] to assign a neuroanatomical label to each native brain voxel. Finally, FreeSurfer 6.0 was used to obtain the volume and the cortical thickness of different areas of interest.

Allelic characterization
APOE polymorphism (haplotype) was determined by an analysis of the genotype of the two polymorphisms (SNPs) that determine it: rs7412 and rs429358. The DNA was extracted from peripheral leukocytes using DNAzol ® Genomic DNA Isolation Reagent (Molecular Research Center, Inc., Cincinnati, OH, USA), following

Statistical analysis
Statistical analysis was carried out in SPSS 25.0 (SPSS Inc., Chicago, IL, USA). Data are reported as the median (interquartile range). The differences between study groups (FH− ApoE ɛ4− and FH− ApoE ɛ4+) in qualitative variables were analyzed using a chi-squared test. We first compared macular area volumes and peripapillary nerve fiber layer thicknesses between the study groups using the Mann-Whitney U test. Secondly, we compared the volumes of brain areas between the study groups also using the Mann-Whitney U test. Finally, partial correlation controlling for age using Pearson's correlation coefficient was applied to study the possible association between macular and peripapillary retinal sectors and brain structures that are mainly related to AD. A colorimetric scale has been applied to the degree of correlation of the variables, where low correlations (r with values between 0.38 and 0.60) have a yellowish color, medium correlations (r with values between 0.60 and 0.72) have orange colors, and strong correlations (r with values between 0.72 and 1.00) have reddish tones. A p-value <0.05 was considered statistically significant.

Demographic data
Demographic data for the FH− ApoE ɛ4− and FH+ ApoE ɛ4+ groups are shown in Table 1.
There were no differences between study groups in terms of age and MMSE (p > 0.05). The FH− ApoE ɛ4− group showed a median age of 60.0 (54.0-64.5) years and the FH+ ApoE ɛ4+ group had a median age of 57.0 (54.0-61.0) years. The median MMSE scores were 29.0 (29.0-29.0) in the FH− ApoE ɛ4− group and 29.0 (28.0-29.0) in the FH+ ApoE ɛ4+ group.
In terms of sex, there were statistically significant differences (p < 0.05) between the numbers of males (11) and females (23) participating in the ApoE ɛ4+ group.

Magnetic resonance imaging
Regarding the cortical thickness, there were no significant differences in any of the regions analyzed between the study groups (p > 0.05) ( Table 2). Regarding the volume of the different brain regions studied, there were no significant differences in any of the regions analyzed between the study groups (p > 0.05) ( Table 3).

Retinal layer volume between groups
Regarding the total retinal volume, there were no significant differences in any of the sectors analyzed between the study groups (p > 0.05).
In the GCL, the FH+ ApoE ɛ4+ group showed a slight, nonsignificant volume reduction of all sectors in comparison to the FH-ApoE ɛ4- (Table 4).
In ONL, EPR, and pRNFL, there were no statistically significant differences (p > 0.05) in any of the sectors analyzed between the study groups (Table 4).

Correlations between the retina and brain in the studied groups
We studied Pearson correlations between the retina (macular sectors and pRNFL) and specific brain structures in each study.
We highlighted correlations of brain MRI and OCT retinal volumes in those retinal sectors where we had found statistically significant retinal volume changes between the FH+ ApoE ɛ4+ and FH− ApoE ɛ4− groups.
The thickness of the temporal sector of pRNFL was only significant correlated with the thickness of the left posterior cingulate region (r = 0.405) and right lingual gyrus (r=0.422) ( Table 6).
The thickness of the inferotemporal sector of pRNFL was significant associated with (i) right and left posterior cingulate thickness (r = 0.539 and r = 0.448, respectively), (ii) right and left isthmus cingulate thickness (r = 0.379 and r=0.434, respectively), and (iii) right ventral diencephalon volume (r = 0.381) ( Table 6 and Fig. 5).
The thickness of the nasal sector of pRNFL was significantly correlated with the right entorhinal volume (r = −0.536) ( Table 6).

Macular sectors and brain correlations
In the IPL, we found an association between the outer macular ring in the nasal sector (N2) and (i) left parahippocampal volume (r = −0.429), (ii) right entorhinal    (Table 7 and Fig. 6).
Finally, in the OPL, the inferior sector (I2) of the outer macular ring was associated with left rostral anterior cingulate thickness (r = −0.506) ( Table 7 and Fig. 6).
The thickness of the superotemporal sector of pRNFL was significantly associated with the (i) right posterior cingulate thickness (r=0.  (Table 8 and Fig. 7).

Discussion
To our knowledge, no previous study has studied the relationships between the retina and different brain structures in cognitively healthy participants but with two main genetic risk factors for developing the disease: (i) having a family history of sporadic senile form of AD and (ii) carrying at least one ɛ4 allele for the allele of the ApoE gene. One of the chief assets of this study was the careful selection of cases. Only family members of people diagnosed with sporadic senile AD were selected, and participants were free of ocular pathology and mental or cognitive disorders that could mask the results. We analyzed both macular and peripapillary areas of the retina in correlation with 20 brain structures, making this one of the studies that analyzed the correlation of the largest number of structures.
The higher number of female participants in the FH+ ApoE ɛ4+ group in comparison with male participants is due to the fact that women are more involved in caring for sick parents [26]. They are also more aware that participating in the studies will further advance the Table 5 Significant age-adjusted correlations between retinal sector volumes and the volumes and thickness of brain structures in participants with high genetic risk of developing AD. A colorimetric scale has been applied to the degree of correlation of the variables knowledge of Alzheimer's disease to help both patients and their descendants [27].
Regarding the brain structural cortical thickness and volume, we did not find any correlation with the risk of developing an Alzheimer's dementia based on familiar history or ApoE ɛ4 phenotype, except for the right cingulate isthmus, which almost reached a statistical difference between both study groups. Although this region is classically involved in AD degeneration, we cannot assess its significance as it is itself affected by all the other Table 6 Significant age-adjusted correlations between pRNFL thickness in different sectors and thickness and volume of brain structures in participants with a high risk of developing AD. A colorimetric scale has been applied to the degree of correlation of the variables FH+ participants with a family history of AD, ApoE ɛ4+ carriers of ApoE ɛ4, pRNFL peripapillary retinal nerve fiber layer, G global, T temporal, ST superotemporal, IT inferotemporal, N nasal, SN superonasal, right right hemisphere, left left hemisphere Fig. 4 Scatter plots of some statistically significant age-adjusted correlations between retinal sector volumes and volumes and thickness of brain structures in participants with high genetic risk of developing AD. mRNFL macular retinal nerve fiber layer, C0 central macular sector, OPL outer plexiform layer, I2 inferior sector of the outer macular ring regions in our research. In the present study, a trend of a larger volume, without reaching statistical significance, was observed in different brain areas in the FH+ ApoE ɛ4+ group compared with the FH− ApoE ɛ4− group.
Classically, it has been reported that subjects carrying at least one E4 allele for ApoE, throughout the temporal continuum of the disease and from several decades before, present a reduction of hippocampal volume [28][29][30][31] or a focal atrophy of this area [32,33]. Nevertheless, an inflammatory reaction mediated by progranulin has been described in patients in early stages of the disease, who already present positive markers for amyloid, which also contributes to producing neuroinflammatory structural changes in preclinical stages of the disease [34]. From another point of view, our group has demonstrated previously that, when compared with the healthy control group, mild cognitive impairment patients exhibited a marked decrease in functional connectivity over posterior areas accompanied by an increased in anteriorventral regions of the brain, representing the common feature of the network failure starting in the pre-dementia stages of the disease, as a compensatory mechanism [35]. Although this increased connectivity has not been shown to require an increase in volume, it remains plausible that the increase in neuronal plasticity required to produce it would carry a transient physical increase in networks structures.
Both considering the compensatory or inflammatory hypothesis, a final possible explanation for this slight not significant increase in volume in our participants could be due to a statistical artifact being necessary to verify it in more extensive and longitudinal studies.
In our study, the FH+ ApoE ɛ4+ group showed a statistically significant volume decrease in the macular area in different retinal sectors compared with the FH− ApoE ɛ4− group. These results are in agreement with our previous study, in which we observed that there is a thinning of certain retinal sectors in relatives at high genetic risk for the development of AD [12].
It has been observed that family history of AD and the ApoE ɛ4 gene were associated with a thinning in the entorhinal cortex, subiculum, and medial temporal lobe, with these factors being additive to each other [18] and these structures the first to show signs of AD [36]. However, in our work, no statistically significant brain changes were observed between participants with FH+ ApoE ɛ4+ and FH− ApoE ɛ4−. This result is supported by previous studies that showed that retinal changes appear earlier than brain changes [13,[37][38][39]. Also, these retinal alterations, which could occur through retrograde transneuronal neurodegeneration [40], may be associated with atrophic brain changes already present before the appearance of clinical cognitive symptoms [17,41]. The relationship between the brain and retina has been analyzed in other populations without a family history of AD and ApoE ɛ4 gene. There are previous studies that analyzed these correlations in normal older adults with mean ages of 68.0 ± 5.3 years, in which pRNFL thickness in the temporal quadrant was associated with temporal medial lobe volume and hippocampus volume, while the inferior quadrant was significantly associated with occipital lobe volume and selectively associated with the substructure of lingual gyrus volume [6]. In our work, in contrast to Shi et al., we did not find a correlation between the inferior quadrants of the pRNFL and the volume of the occipital lobe.
In a recent work undertaken in an older population (mean age: 65.1 ± 9.0 years), a positive correlation was observed between the pRNFL thickness and right and left hippocampal thickness [42]. Also, Méndez-Gómez et al., in an older population of 80.8 ± 3.9 years, found a direct correlation between pRNFL and the hippocampal fraction [43]. In agreement with these authors, we found a direct correlation in the FH+ ApoE ɛ4+ group between the thickness of the inferotemporal region of the pRNFL with (i) the thickness of right and left posterior cingulate; (ii) the right and left isthmus cingulate, being these, hippocampal areas; and (iii) the volume of right ventral diencephalon. The association between the retina and medial temporal lobe volume may indicate a degeneration in both tissues at the same time, preceding clinical cognitive changes in cognitively healthy participants at risk of AD [17].
In a 12-month longitudinal study, with participants with a mean age of 71.8 ± 3.9 years, an inverse association between the mean reduction in pRNFL thickness and the decrease in central cingulate cortex volume was observed. In addition, the reduction of pRNFL thickness in the inferior quadrants was associated with the decrease of central cingulate cortex volume [44]. We found the same correlation between inferotemporal pRNFL thickness with posterior cingulate thickness in our FH+ ApoE ɛ4+ participants. One possible explanation of the differences between previous studies and our work could be that our participants are young elder people with a mean age of 58.8 ± 6.0 years and we only accepted MMSE values > 26, whereas in Shi and colleagues' studies, they accepted values for the Chinese version of the Mini-Mental State Examination (CMMSE) of ≥24. This issue is important, because normal cognitive values are above 26, and lower values could mask previous stages of the disease such as subjective memory complaints or mild cognitive impairment. Furthermore, we found no significant differences in pRNFL thickness between our study groups. Altered pRNFL thickness is known to be a good marker of disease progression [16]. On the other hand, changes in this retinal layer are associated with increased susceptibility to accumulation of neurofibrillary tangles and deposition of amyloid plaques in the occipital lobe and inferior temporal lobe, which are a part of the visual association cortex [45,46]. Ong et al. observed, in a group of 164 patients aged 40-85 years including patients with cognitive impairment without dementia (n = 125), cognitively healthy participants (n = 36), and people with dementia (n = 3), that GCL-IPL thinning is associated with a reduction of gray matter in occipital and temporal lobe volume [47]. In another study performed on 79 neurologically normal adults with a mean age of 76.0 ± 5.5 years, the pRNFL reduction and the decline in total macular volume and GCL volume were significantly associated with a decrease in the medial temporal volumes of the hippocampus, the parahippocampal region, and the entorhinal region [17]. In a study carried out by Mutlu and colleagues, as part of the Rotterdam Study, older patients (67 ± 9.8 years old) with MMSE values of 28 ± 1.7, only the data of GCL, IPL, and pRNFL thickness measurements in relation to brain volume and hippocampal volume were analyzed [48]. In this study, it was found that the thinning of these retinal layers was significantly associated with lower brain volume and lower hippocampal volume, finding that this retinal thinning was associated with the thinning of both gray and white matter in the brain [48]. Chua et al., in a large population-based study (2131 participants) aged between 40 and 69 years old, also found that GCL thinning and total macular thickness were significantly associated with smaller hippocampal volume [49]. These findings are consistent with another study in 20 cognitively healthy, younger patients (50.5 ± 14.1 years old), where it was found that entorhinal gray matter volume was correlated with the GC-IPL complex [50]. In our FH− ApoE ɛ4− group, there was a significantly inverse correlation in the IPL between the volumes of the outer nasal macular sector and the volumes of the left parahippocampal region, the right entorhinal region, the right and left medial temporal lobule, the right and left hippocampus, the right lingual gyrus, the left cuneus, the right ventral diencephalon, and the left amygdala (Table 7). In addition, there was a significant inverse correlation in this retinal layer between the outer inferior macular sector and the right entorhinal, right medial temporal lobe, right ventral diencephalon, right hippocampus, and right and left amygdala ( Table 7).
Carriers of the ApoE ɛ4 gene have a thinning of the entorhinal cortex in the left hemisphere compared with the right hemisphere, regardless of age or cognitive status [51]. In our FH− ApoE ɛ4− participants, there was a significant correlation between the left entorhinal thickness with the C0 retinal volume in the INL. Asymmetry of cerebral cortex thickness in the entorhinal region associate to the presence of an ApoE ɛ4 allele [52], may in itself reflect pathology, but could lead to future spatial and temporal distribution patterns of pathological changes [51]. This fact may explain, why in our participants with ApoE ɛ4+, the correlation between the nasal sector of the pRNFL with the right entorhinal volume was stronger than in ApoE ɛ4participants.
As can be seen in the tables, most of the correlations found between the macular sectors and brain structures in FH− ApoE ɛ4− participants are inverse, while in participants with a high genetic risk for sporadic senile form of AD (FH+ ApoE ɛ4+), they were mostly direct. This could be due to the presence or not of the ApoE ɛ4 gene, as we have already seen in the retina [12], as well as to different behaviors in the central nervous system generated by the presence of this gene [51]. When the ApoE ɛ4 gene is present, changes could occur at the same time in the retina and in different brain areas, either due to the accumulation of Aβ or tau protein, or to structural brain changes that would be concomitant with retinal changes. However, the absence of the ApoE ɛ4 gene could generate different behavior as the correlations are inverse and a change in brain structures would not be accompanied by retinal changes, because they would not be related to it or would not be concomitant in their genesis.

Limitations
Our study has strengths and limitations. The main strength of our work is that it is one of the brain-retina correlational studies that has analyzed the most brain areas, considering both right and left hemispheres. One of the limitations of our study is that our sample is small; however, we have made a strict selection of our participants: all participants at high genetic risk have a family history of AD and at least one ɛ4 allele for ApoE. In addition, all participants were cognitively healthy and had a MMSE score above 26. Furthermore, despite the small number of participants, the OCT results are already consistent, so if the sample was increased, certain MRI results, which are close to significance, would also be statistically significant. Longitudinal studies with larger samples of participants should be performed to confirm the etiopathogenic mechanisms involved in the changes occurring between the retina and the brain. Such studies are necessary to understand the evolution of imaging biomarkers as well as to better predict the possible establishment of the disease in participants at high genetic risk for the development of AD, so participants with mild cognitive impairment or more advanced stages of the disease should also be included. Furthermore, if our results were replicated by other groups, it would strengthen the interpretation of our findings on retina-brain correlations. In future studies, it would be necessary to include other biomarkers for AD such as cerebrospinal fluid analysis or positron emission tomography as they would increase the value of our findings [42].

Conclusions
In conclusion, these results demonstrate that there is a correlation between changes in the retina and various brain structures in participants at high genetic risk for developing sporadic senile forms of AD. In these cognitively healthy participants, there is already a significant correlation between pRNFL thickness and the volume of brain areas closely related to AD such as the entorhinal cortex, the lingual gyrus, and the hippocampus. Moreover, with the recent approval by the Food and Drug Administration (FDA) of the first treatment capable of modifying the pathophysiology of AD, the search for cheap, non-invasive, and readily available biomarkers will be mandatory given the need for early diagnosis in participants at high risk of a pathology whose incidence will increase exponentially in the near future worldwide [53]. Therefore, OCT volume measurements and their correlations with brain area volumes could be a biomarker of AD, even in the preclinical stages of AD, and longitudinal studies are needed to really know how many of these participants eventually develop the disease.