*2.3. Statistical Analysis*

The data analysis was conducted using SPSS version 22.0 (SPSS, Inc., Chicago, IL, USA). As there were no previous similar studies, no sample size calculation was performed in this investigation. The results of the right and left eyes were separately compared between the groups. Gender was compared between the groups by a Fisher's exact test. The quantitative variables were assessed for normality distribution by inspecting histograms and using the Shapiro–Wilk test. Normally distributed variables were expressed as the mean and standard deviation, while non-normally distributed values were expressed as the median and interquartile range. Comparisons between two normally distributed variables were performed with the unpaired Student's t-test. If at least one variable was non-normally distributed, the comparison between the groups was made by the Mann–Whitney test. A correction for multiple comparisons was not applied to this study in order to avoid the false-negative results [21]. A *p*-value less than 0.05 was considered statistically significant.

#### **3. Results**

The OCT examinations of one right eye from the dyslexic group and two left eyes from the control group were discarded due to segmentation errors at one or more sectors. Finally, 89 reliable OCT scans from 46 participants were selected in this study: 24 right eyes (7 men, 17 women) and 25 left eyes (8 men, 17 women) were selected from 25 normal controls. Moreover, 21 right eyes (7 men, 14 women) and 19 left eyes (5 men, 14 women) were included from 21 dyslexic subjects. The gender between the dyslexic and control groups did not differ when considering the right (*p* = 1, Fisher's exact test) or the left eyes (*p* = 0.749, Fisher's exact test).

Similarly, the mean age was not different between the groups for the right eyes (15.83 ± 3.81 years for dyslexics, with a range of 9 to 23 years, and 16.00 ± 4.11 for normal controls, with a range of 10 to 23 years, *p* = 0.889, unpaired Student's *t*-test) or for the left eyes (15.76 ± 3.75 years for dyslexics, with a range of 10 to 23 years, and 16.26 ± 3.97 for normal controls, with a range of 10 to 23 years, *p* = 0.672, unpaired Student's *t*-test).

The thicknesses in all four parafoveal sectors were significantly thicker in both the right and left eyes of the dyslexic group in the following segmentations: complete retina (Table S1), inner retina (Table S2), middle retina (INL + OPL + ONL) (Table S3) and OPL + ONL

**Figure 2.** Schematic heatmaps of the statistically significant thickenings (expressed as microns) in the dyslexic group compared to the normal control group for the right and left eyes (mean differences in thickness) and for different segmentations. See also Figure 1a–c. White sectors indicate no significant differences. IPL = inner plexiform layer, INL = inner nuclear layer, OPL = outer plexiform layer, ONL = outer nuclear layer. See also Tables S1–S4 and S7.

In contrast, no thickness differences were observed between both the groups in any of the sectors of the retinal nerve fiber layer (RNFL) (Table S8), ganglion cell layer (GCL) (Table S9), ganglion cell complex (RNFL + GCL + IPL) (Table S10) or outer retina (Table S11) in either eye.

When considering the inner ring subfield (parafovea), we found that the complete retina, inner retina, middle retina (INL + OPL + ONL), OPL + ONL, IPL + INL and INL showed a higher thickness in the dyslexic group for both eyes, with IPL showing this difference only for the right eye (Table 1). The rest of the segmentations (outer retina, GCC, RNFL, GCL) did not present any significant differences in either eye (Table 1).

**Table 1.** Thickness comparison between the groups for the inner ring subfield (parafovea) in the ETDRS grid.


Right and left eyes were independently compared between the groups. The thickness results are expressed as microns. Statistically significant results are depicted in bold. Std. Dev = standard deviation, Dif. = difference, UTT = unpaired *t*-test, GCC = ganglion cell complex, RNFL = retinal nerve fiber layer, GCL = ganglion cell layer, IPL = inner plexiform layer, INL = inner nuclear layer, OPL = outer plexiform layer, ONL = outer nuclear layer.

In contrast, when dealing with the outer ring subfield (perifovea), only INL + OPL + ONL and INL showed a discrete increase of thickness in the dyslexia group, only when comparing the right eyes (Table 2).


**Table 2.** Thickness comparison between the groups for the outer ring subfield (perifovea) in the ETDRS grid.

Right and left eyes were independently compared between the groups. The thickness results are expressed as microns. Statistically significant results are depicted in bold. Std. Dev. = standard deviation, Dif. = difference, UTT = unpaired *t*-test, GCC = ganglion cell complex, RNFL = retinal nerve fiber layer, GCL = ganglion cell layer, IPL = inner plexiform layer, INL = inner nuclear layer, OPL = outer plexiform layer, ONL = outer nuclear layer.

### **4. Discussion**

This study found parafoveal thickenings in both eyes of the dyslexic group for the middle retina, mainly the ONL-related segmentations, but also for the INL-related segmentations (Table 1). Remarkably, no difference was noted in the GCL-related segmentations or in the outer retina of right or left eyes. Furthermore, complete retina and inner retinal thickenings found at the parafovea in this study seem to be due to the middle retinal thickenings (Figure 1a; see also Table 1 and Tables S1–S4 and S7). Additionally, a foveal thickening was also observed in ONL + OPL (in both eyes) and INL + ONL + OPL (in the right eye).

The ONL contains photoreceptor cell bodies and the INL bipolar, amacrine and horizontal cells nuclei. OPL and IPL are made up of axons and synapses and connect the neighboring nuclear layers (ONL-INL and INL-GCL, respectively) [9]. In fact, the ONL and INL layers share a common embryological origin: the outer neuroblastic zone differentiates into ONL and INL after fetal week 10. Then, first synapses appear in the IPL and OPL by fetal week 12 [10]. ONL and INL the somata are displaced during normal foveal development (from fetal week 22 to postpartum month 45): photoreceptor cell bodies (ONL) present a centripetal displacement toward the foveal center with cellular packing and elongation, while INL and GCL have centrifugal displacement to the foveal rim [10]. One possible explanation for the thickenings found in this study is a disorder in foveal development due to a gap between these two movements in the opposite direction. In this way, orientation of the somata and the axons of ONL- and INL-related cells could result in more vertical increasing of the thickness of these layers. This mechanism has been proposed to explain some similar retinal features found in the eyes of preterm patients with

foveal immaturity [22]. Other alternative explanations for the found thickenings could be a neuronal/glial population increase, cell size augmentation or an extracellular expansion in the affected segmentations. More studies are warranted to elucidate this question.

A genetically determined disorder in foveal development could bring about the findings of the present study. This hypothesis is consistent with the fact that dyslexia presents a high degree of heritability (70% or even more), whereas the environment has little effect [23–25].

As a matter of fact, some genetic polymorphisms have been detected to be much more prevalent in dyslexia. These polymorphisms are in relation to abnormal axonal growth and defective neural migration [26–28], and have been associated with alterations in the development of cortico-cortical and cortico-thalamic circuits in dyslexia [2,29], so they may similarly lead to the retinal differences detected in this study.

The highest thickenings found in this study were located at the parafovea, although alterations at the fovea, and even at some perifoveal sectors, were also present in OPL + ONL and INL + OPL + ONL segmentations (Figure 2). The fovea captures the visual field one degree around the fixation point, while the parafovea is surrounding the fovea up to five degrees from the fixation point [30,31].

The fovea and the parafovea work together in the reading process, because, although the fovea is oriented to the target word, the parafovea previews the next words to facilitate further foveal processing [32,33]. In fact, parafoveal recognition of embedded letters and words has been proven to be worse in dyslexic subjects than in normal controls [34–36]. Furthermore, reduced and delayed parafoveal preview benefits have been associated with dyslexia [37–40].

Thus, the thickenings found in this study could constitute the morphological correlation of parafoveal dysfunction in dyslexia. These thickenings may theoretically cause a higher level of light scattering transmitted to the outer segments of photoreceptors, where the phototransduction takes place, and potentially a subsequent loss of sensitivity [10]. The other possibility is that the found thickenings could be related to a change in the arrangement of the photoreceptors and/or Müller cells, as suggested above. An incorrect arrangement of these cells may induce an incorrect angle of incidence of light and a subsequent reduction of sensitivity due to the Stiles–Crawford effect of the first kind [41,42]. These, or other causes associated to these thickenings, may alter parafoveal preview function. Further study is required in this sense.

It is also remarkable that the thickening in all four parafoveal sectors (360 degrees) found herein is consistent with the fact that dyslexia is present in all languages, independently of the reading direction going from left to right (English, Spanish, French, German), from right to left (Hebrew or Arabic) or from top to bottom (Japanese, Chinese) [6].

This exploratory research also has its limitations. First, the sample size is small and the outcomes should be studied in larger groups, but we found significant results. Moreover, the facts that the comparisons have been calculated separately for the right and the left eyes and that the found differences are so similar in both eyes, affecting the same segmentations and with analogous disparities, reinforce the validity of our results (Figure 2). Second, the participants included in this study are mainly adolescents and young adults, so we cannot extrapolate our results to other age groups. Third, this study is cross-sectional, so we cannot know whether the found differences are stable over time and whether reading interventions are able to remodel retinal structures, similarly to what has been found in other investigations on the CNS in dyslexia [7,8]. Fourth, the grouping of this study is based on a diagnosis and not on specific reading measurements. Using a categorical method to create the groups does not permit to explore a quantitative correlation between reading measurements (for example, reading speed) and OCT parameters. However, we have to keep in mind that this is an exploratory investigation. Further works should deal with these relationships. Fifth, it should be pointed out that the nomenclature of the inner and outer retina, as defined by the OCT device used in this study (Spectralis, Heidelberg Engineering), is inexact (Figure 1). The inner and outer retina are major divisions in both neurobiology

and vascular biology. The outer retina includes the OPL, ONL, photoreceptors and retinal pigment epithelium, and receives blood supply from choroidal circulation, whereas the inner retinal layers are dependent on retinal circulation [43]. However, this fact does not affect the results of this study. Sixth, this study is limited in analyzing morphological results. Further functional studies (i.e., mERG) in relation to morphological differences should be investigated in further studies. Seventh, the OCT examinations were only analyzed by one expert, so inter-variability agreement cannot be assessed. However, all the examinations fulfilled the reliability criteria. Finally, this study is limited to describe morphological differences in the macula of dyslexic subjects, so we cannot determine if these findings are causes or consequences, or the parallel manifestations in the pathophysiology of dyslexia: if these morphological findings were the cause, dyslexia could be primarily a retinal disorder, as suggested above. If they were the consequence or a parallel manifestation, dyslexia would be capable of remodeling retinal structures, as is done in other CNS structures [2,6,44]. Thus, the macula could be a privileged and accessible site to study dyslexia by using a fast, inexpensive and non-invasive technique, such as OCT. Further studies are required in this sense. Nevertheless, our significant results open a new horizon for the investigation of dyslexia.

## **5. Conclusions**

From this exploratory research, we conclude that the macular morphology differs in dyslexic and normal controls, especially in the parafovea.

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/jcm12062356/s1, Video S1 and Tables S1–S11 have been added as supplementary material. Video S1. Three-dimensional (3D) video that dynamically combines the cross-sectional and the en face OCT scans of the macula from a right eye. The cross-sectional scans that configure the 3D surface of the macula progressively disappear, while the two-dimensional image of the en face scan remains. See also Figure 1a,b.

**Author Contributions:** Conceptualization, J.J.G.-M.; Methodology, J.J.G.-M., P.S.-C., C.G.-M., M.D.-P.- M., V.Z.-M., M.D.P.-D. and M.D.-R.-V.; Formal analysis, J.J.G.-M., N.B.-M., C.G.-M., E.R.-V., M.D.-P.-M., V.Z.-M., M.D.P.-D. and M.D.-R.-V.; Investigation, P.S.-C., C.G.-M., E.R.-V., M.D.P.-D. and M.D.-R.- V.; Data curation, J.J.G.-M., N.B.-M. and P.S.-C.; Writing—original draft, J.J.G.-M., N.B.-M., P.S.-C., C.G.-M., E.R.-V., M.D.-P.-M., V.Z.-M., M.D.P.-D. and M.D.-R.-V.; Writing—review & editing, J.J.G.-M., N.B.-M., P.S.-C., C.G.-M., E.R.-V., M.D.-P.-M., V.Z.-M., M.D.P.-D. and M.D.-R.-V.; Supervision, J.J.G.-M. and M.D.-R.-V. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** The study was approved by the Local Ethics Committee at the University General Reina Sofia Hospital of Murcia, Spain (protocol 005/2017), and adhered to the Declaration of Helsinki criteria.

**Informed Consent Statement:** The participants or their legal representative signed an informed consent form.

**Data Availability Statement:** The data sets generated and/or analyzed during this study are available from the corresponding author on reasonable request.

**Acknowledgments:** This work has been done in part with the collaboration of members assigned to the research team of Murcia and Valencia, pertaining to the ophthalmology network OFTARED (RD16- 0008), as well as the members of the inflammatory diseases network RICORS (RD21/0002/0032) of the Institute of Health Carlos III (ISCIII), Spanish Government (Madrid, Spain). We would like to thank Jose Manuel Tamarit (Heidelberg Engineering, Heidelberg, Germany) for his technical help with the OCT device and Guadalupe Ruiz (Statistics Department, FFIS-IMIB, Murcia, Spain) for her statistical assistance in the data analysis of this study. The preliminary results of this study were presented during the Association for Research in Vision and Ophthalmology (ARVO) Annual Meeting, originally scheduled for 2–6 May in San Francisco, CA, USA, 2021, but finally presented in a

Virtual Meeting. The final results of this study were presented at the 23rd Congress of the European Association for Vision and Eye Research (EVER), 13–15 October 2022 in Valencia, Spain.

**Conflicts of Interest:** The authors have no relevant financial or non-financial interests to disclose.
