Optical coherence tomography as a potential surrogate marker of dopaminergic modulation across the life span

The retina has been considered a "window to the brain" and shares similar innervation by the dopaminergic system with the cortex in terms of an unequal distribution of D1 and D2 receptors. Here, we provide a comprehensive overview that Optical Coherence Tomography (OCT), a non-invasive imaging technique, which provides an "in vivo" representation of the retina, shows promise to be used as a surrogate marker of dopaminergic neuromodulation in cognition. Overall, most evidence supports reduced retinal thickness in individuals with dopaminergic dysregulation (e.g., patients with Parkinson ’ s Disease, non-demented older adults) and with poor cognitive functioning. By using the theoretical framework of metacontrol, we derive hypotheses that retinal thinning associated to decreased dopamine (DA) levels affecting D1 families, might lead to a decrease in the signal-to-noise ratio (SNR) affecting cognitive persistence (depending on D1-modulated DA activity) but not cognitive flexibility (depending on D2-modulated DA activity). We argue that the use of OCT parameters might not only be an insightful for cognitive neuroscience research, but also a potentially effective tool for individualized medicine with a focus on cognition. As our society progressively ages in the forthcoming years and decades, the preservation of cognitive abilities and promoting healthy aging will hold of crucial significance. OCT has the potential to function as a swift, non-invasive, and economical method for promptly recognizing individuals with a heightened vulnerability to cognitive deterioration throughout all stages of life.


Introduction
Optical Coherence Tomography (OCT) is a non-invasive imaging technique allowing to capture the retinal layer similar to an "in vivo" histology.The quantitative measurement provides the thickness of the retinal nerve fiber layer (RNFL), but also the macula thickness and volumetric measures (e.g., macula volume).The retina was often referred to as the "window to the brain", because it develops embryologically from the anterior neural plate or the diencephalon (Graw, 2010;London et al., 2013).Moreover, neurons in the retina and the central nervous system (CNS) share similar features, such as the structure of the cells with soma, axons and dendrites (for a review, see London et al., 2013) or the specific slow, rhythmic firing pattern of retinal dopamine (DA) cells found in DA neurons of the substantia nigra and the ventral tegmental area (Hirasawa et al., 2015).
Previous reviews mainly focused on OCT as a potential diagnostic tool for mental disorders such as schizophrenia or unipolar depressive disorders (Adams and Nasrallah, 2018;Duraković et al., 2020;Schönfeldt-Lecuona et al., 2014;Tan et al., 2020), and neurodegenerative diseases such as Parkinson's disease (PD) (Jee-Young Lee et al., 2014;Mailankody et al., 2019) and Alzheimer's disease (AD) (Cabrera DeBuc et al., 2019).In contrast, the aim of this perspective article is to provide a comprehensive synthesis of the literature on OCT and its relation to dopaminergic neuromodulation and cognition across the lifespan in order to develop a "surrogate marker hypothesis of OCT" that is relevant to capture ageing-related processes from a lifespan perspective.We rely on the fact that a) extensive research supported by pharmacological challenges and individuals suffering from dopaminergic abnormalities indicate that OCT is closely related to dopaminergic function and that b) DA is directly involved in regulating cognitive processes (Nieoullon, 2002).We will further substantiate our line of reasoning by taking the theoretical framework of metacontrol into account (Hommel and Colzato, 2017).
We place great importance on this approach due to the criticality of intact cognition throughout our life span, playing a vital role in educational and work success (Diamond, 2013) as well as healthy aging (Rowe and Kahn, 1997).As we grow older, cognitive processes experience a decline (Harada et al., 2013).Nevertheless, older individuals who maintain intact cognition are more adept at managing their daily activities, staying engaged with their surroundings, and effectively communicating with others, ultimately leading to a heightened sense of self-efficacy and overall life satisfaction (Rowe and Kahn, 1997).It is important to note that cognitive resilience may indeed help elderly individuals better cope with the typical cognitive changes that occur with age and potentially reduce their susceptibility to severe cognitive impairments (Rowe and Kahn, 1997).Considering the substantial decrements in cognition that accompany human aging (Harada et al., 2013) and the rapid aging of the global population (Powell, 2010), the search for new biomarkers to predict cognitive impairment becomes crucial.Enhancing the ability to diagnose cognitive decline more accurately can significantly contribute to the improved management of cognitive aging (Schneider and Yvon, 2013).Therefore, the discovery and implementation of these biomarkers hold great promise in facilitating early detection and intervention for cognitive impairment across the life span.
In the following, we start with a brief historical overview of OCT.In Section 3, we will outline the anatomy of dopaminergic supplies in the retina and how OCT can measure them based on pharmacological challenges of OCT in animals and humans.Moreover, we will review baseline OCT measurements in individuals affected by diseases in which the dopaminergic system is dysregulated.Following, in Section 4, we will summarize studies that examine the relationship between cognitive function and OCT parameters in patients with dopaminergic abnormalities.Finally, we will propose OCT as a potential surrogate marker of dopaminergic neuromodulation and associated cognition in forecasting cognitive deterioration in ageing from a lifespan perspective by addressing current challenges and future research directions in Section 5.

Historical background
The beginnings of OCT date back to the 1980 s.At that time, various groups were investigating white light interferometryan optical measurement method that uses the interference of broadband light to measure the three-dimensional structures of an object or tissue.In the early 1990 s, Huang et al. (1991) then coined "Optical Coherence Tomography".The functional principle is as follows: In OCT, light waves are directed at the tissue to be examined.The tissue structure reflects these waves, and their delay is measured to determine the depth at which the reflection took place (for further details, see Aumann et al., 2019).Since its first application, the technological capabilities of OCT have continued to improve, allowing higher resolution and faster image acquisition rates by spectral-domain or swept source OCT and higher frequencies.Due to the ease of use, OCT is widely used in ophthalmology to diagnose eye diseases and has become a popular method to study retinal markers for various diseases, among others, underlying dopaminergic dysregulations (e.g., PD).Therefore, OCT was primarily used as a diagnostic tool in clinical settings.However, since all retinal layers contain dopaminergic cells (Roy and Field, 2019), we propose that OCT should be considered as a potential tool for predicting dopaminergic function in cognitive neuroscience and its role in ageing from a lifespan perspective.We suggest that OCT might be valuable to further investigate the neuromodulation of cognitive processes related to DA, the main neurotransmitter likely indexed by this method, in healthy humans or patients.In the past, other methods have been proposed to quantify DA in an indirect, non-invasive and cost-effective way such as eye blink rate (e.g.Jongkees and Colzato, 2016), electroretinogram (Lavoie et al., 2014;Nasser et al., 2013;Schwitzer et al., 2017), or visual evoked potentials (Nguyen et al., 2017).However, unlike eye blink rate, visual evoked potentials or electroretinogram, which measure functional correlates of DA-associated parameters and behavioral outcomes, OCT is an imaging technique that obtains high-resolution structural images (at the micrometer scale) of biological tissue (e.g.layers of the retina) and is thus a more direct tool to approach dopaminergic function with a clear neuroanatomical basis.Compared to Positron Emission Tomography (PET), which is the only method to study DA concentration in the brain in vivo and requires a radioactive tracer's expensive, invasive use, OCT is a non-invasive, low-cost, rapid, and easy to administer examination.To support our claim, in the next section we will examine the relationship between retinal OCT parameters and dopaminergic function in the brain and describe the effects of pharmacological influences on DA in animals and humans.

OCT and dopaminergic function
In this section, we will outline the anatomy of the dopaminergic supply in the retina and show how OCT can measure this dopaminergic supply using pharmacological challenges acting on DA.
The retina is composed of several layers of different types of neurons connected by synaptic transmission.Here, the retina can be divided into the inner retina with the RNFL, ganglion cell layer (GCL), inner plexiform layer (IPL), inner nuclear layer (INL), outer plexiform layer (OPL), and outer nuclear layer (ONL), and the outer retina with the photoreceptor layers up to the choroid (see Fig. 1).The axons of the retinal ganglion cells (RGCs) form the RNFL, leave the eye at the optic disc and then form the optic nerve.As the axons of the RNFL are not myelinated, the thickness of this layer is thought to reflect axonal degeneration (Alves et al., 2018).The cell bodies of RGCs are located in the GCL, while their dendrites synapse with bipolar and amacrine cells in the IPL (Vlasits et al., 2019).The cell bodies of dopaminergic amacrine cells (DACs) are part of the INL as are bipolar and horizontal cells (Graw, 2010;Hoon et al., 2014).

Dopaminergic supply in the retina and the relationship between brain dopaminergic function and retinal parameters
DACs are the primary source of DA in the retina (Polli et al., 2021;Popova, 2020;Roy and Field, 2019).All retinal layers express DA receptors, but the distribution of different DA receptor types varies.While D1 receptors are mainly expressed by DACs, horizontal and bipolar cells in the inner retina, rods and cones in the outer retina express D2-like receptors (i.e., mainly D4 and probably D2 in some species).RGCs, which are also part of the inner retina, express both D1 and D2 receptors, although D2 receptors may be more sparsely expressed (Roy and Field, 2019).D1 receptors are thus predominantly present in the retina, as is the case in the prefrontal cortex (PFC) (Durstewitz and Seamans, 2002).Since research on dopaminergic receptor types in the retina is still ongoing, we propose to use the very layers of the retina that contain D1 and D2 receptors as possible surrogate markers of dopaminergic function as indexed by retinal OCT parameters (i.e., GCL, IPL and INL).
Several studies indicate a relationship between brain dopaminergic function and retinal parameters.Most of these findings are based on research into dopaminergic drugs in patients with dopamine deficiency or excess, as well as experimental stimulation or blocking of dopaminergic receptors in healthy volunteers (for a detailed review, see Brandies and Yehuda, 2008).For example, DA antagonists used to treat schizophrenia (Nguyen et al., 2017) and methylphenidate (a DA agonist) in patients with Attention Deficit Hyperactivity Disorder (ADHD) (Werner et al., 2020) showed significant effects in the electroretinogram.Complementing these findings, some researchers noted that antipsychotic medications, which act as DA antagonists, might have toxic effects on the retina.For example, functional changes like reduced contrast sensitivity or abnormal visual evoked potential latencies have been observed after administration of antipsychotics.Blockade of DA receptors could lead to a loss of retinal cells due to reduced activity (for reviews, see Brandies and Yehuda, 2008;Silverstein et al., 2020;Silverstein and Rosen, 2015).
Nevertheless, most knowledge in this area comes from Parkinson's Disease (PD) research, as this disease is characterized by a loss of dopaminergic cells in the brain and retina (Garcia-Martin et al., 2014;Guo et al., 2018).An early postmortem study reported a significantly decreased retinal DA content in PD patients without dopaminergic treatment, whereby patients who received dopaminergic treatment showed a similar DA content as controls (Harnois and Di Paolo, 1990).Further, studies of PD in humans and rodents showed apoptosis of retinal dopaminergic neurons (Ahn et al., 2018;Lee et al., 2020;Tatton et al., 1990) even before a dopaminergic loss in the brain (Normando et al., 2016;Prudêncio et al., 2002).Pharmacological studies underscore the causal relationship between dopaminergic transmission and functional changes in PD (e.g.visual impairment), which significantly improve with DA agonist administration, while DA antagonists worsen impairments (Meoni et al., 2020).Ghilardi et al. (1988) observed decreased DA levels accompanied by functional retinal changes in visual evoked potentials and electroretinogram parameters in rodents after administration of MPTP, a drug known to induce parkinsonism by destroying dopaminergic neurons in the substantia nigra.Studies on healthy individuals are consistent with these findings and show changes in retinal parameters after administration of DA antagonists, e.g.reduced latencies and amplitudes of electroretinogram parameters or delayed visual evoked potentials (for a review, see Brandies and Yehuda, 2008).
As far as OCT is concerned, there are a few studies in animals and patients that specifically examine dopaminergic functions as a primary outcome, but which could confirm previous findings.To our knowledge, only one study has addressed the pharmacological challenges of OCT in animals.Normando et al. (2016) were able to detect retinal changes in a rodent model of PD following administration of rotenone (an insecticide known to destroy dopaminergic neurons; Esteve-Rudd et al., 2011).They found that mean retinal thickness, the RNFL, and the inner and outer photoreceptor segments swelled on day 20 after rotenone-induction.Specifically, the results showed apoptosis of RGCs at day 20 and histological neurodegenerative changes in the substantia nigra and striatum at day 60.According to the authors, one reason for the simultaneous occurrence of RGC apoptosis and retinal swelling after 20 days could be rotenone-induced inflammatory processes.Other explanations for the observed retinal thickening could be increased mitochondrial biogenesis or stagnation of axonal transport preceding RGC loss.These findings highlight that retinal changes may precede the pathological manifestations of PD in the brain and, therefore, serve as early DA markers of disease progression.The group also investigated the neuroprotective effect of rosiglitazone, an antidiabetic drug, that has already shown promising therapeutic results in animal models of PD before.It promotes the inhibition of monoamine oxidase and, thus, reduces the enzymatic degradation of DA.
In humans, few studies have investigated the effects of dopaminergic pharmacological challenges in PD with OCT.Ahn et al. (2018) examined the relationship between retinal OCT parameters (i.e., RNFL, GCL and IPL) and DA transporter densities in the putamen and substantia nigra in 49 drug-naïve patients with PD, suggesting that retinal thinning is associated with dopaminergic loss in the left substantia nigra and correlates with disease severity.The findings from Lee et al. (2020) support this observation emphasizing an association between retinal thinning and DA transporter binding in the striatum in 30 individuals with idiopathic rapid eye movement sleep behavior disorder, a prodromal sign of PD.Moreover, dopaminergic agents such as levodopa may have protective features when looking at retinal thinning (Sen et al., 2014;Sevim et al., 2018;Yang et al., 2016).Confirmatory results came from Yavas et al., (2007), who used Heidelberg Retina Tomograph (HRT), a confocal scanning laser ophthalmoscope utilizing a diode laser source with a wavelength of 670 nm.When capturing the height values that fall on the edge of the optic disc margin, they found no significant difference in RNFL thickness between healthy controls (N = 21) and PD patients treated with levodopa (N = 16), while PD patients administered DA agonists (N = 28) showed thinnest RNFL compared to healthy controls and the levodopa group.The authors explain this counter-intuitive finding (i.e., no observed neuroprotective effect of DA agonists on RNFL) with the fact that the efficacy of DA agonists on DA receptors is lower than the efficacy of DA itself.Given that levodopa is a precursor of DA, it has an efficacy of 100% for all DA receptor subtypes, therefore probably enhancing neuroprotective effects on RNFL compared to DA agonists.One major drawback of the reported findings is their focus on the indirect RNFL estimates, thereby neglecting other dopamine-associated layers, such as GCL or INL.
As shown before, the studies reviewed indicate a robust relationship between dopaminergic functions in the brain and retinal parameters.From a theoretical perspective, the most valid OCT markers for indexing DA activity are the GCL, IPL and INL.Although RNFL thinning is not reported as one of the most valuable marker for DA activity, it might still be, as it should be considered that axonal degeneration may precede the loss of cell bodies (Fricker et al., 2018;Jae-Young Lee et al., 2014;Salvadores et al., 2017;Wang et al., 2012).Therefore, RNFL thinning might be a precursor of RGC degeneration due to retrograde trans-synaptic degeneration originating in the visual pathway (Jindahra et al., 2009;Keller et al., 2014), modulated by DA (Hurd et al., 2001) and closely associated with DA-rich basal ganglia (Anderson, 2019;Kravitz et al., 2013).Hence, RNFL thinning could be a sign of dopaminergic dysfunction.
Indeed, findings from diseases characterized by DA deficiency, like PD (Cools, 2006) and Huntington's disease (Cepeda et al., 2014), indicate a thinning of the peripapillary RNFL (pRNFL) and layers associated with ganglion cells (i.e., GCL, GCL-IPL, and GCC).Whereas for PD, the thinning regarded the whole portion of pRNFL and the RGCs (Chrysou et al., 2019;Yang et al., 2016;Yu et al., 2014;Zhou et al., 2021), notably, for Huntington's disease, most studies found significant differences only in single quadrants (mainly the temporal quadrant) of the pRNFL (Gatto et al., 2018;Kersten et al., 2015;Sevim et al., 2019).The authors interpreted these observations as a consequence of a predominant degeneration of small parvocellular axons due to the impairment of mitochondrial trafficking.However, one study revealed an overall decrease in pRNFL in patients with Huntington's disease (Svetozarskiy et al., 2020), although this was also characterized by a specific parvocellular pattern of retinal neurodegeneration (i.e., predominantly thinning of temporal pRNFL).It should be noted that, in contrast to previous studies, the disease duration of the Huntington's disease patients appeared to be considerably higher in the latter study.This suggests that, although the reduction of RNFL begins in the parvocellular pathway (i.e.within relatively small cells and their thin axons), axonal degeneration also manifests in the magnocellular pathway during the disease.This reasoning is supported by Kersten et al. (2015), who reported a correlation between the decrease in temporal pRNFL and disease duration.Along these lines, Gatto et al. (2018) found significant thinning of the mean total pRNFL in a sub-analysis of patients in more severe stages of Huntington's disease.
In disorders characterized by overactivity of the DA innervated nigral-striatal circuit, such as ADHD (Levy and Swanson, 2001;Tripp and Wickens, 2009) and schizophrenia (Davis et al., 1991;McCutcheon et al., 2018;Sonnenschein et al., 2020), thinning of the pRNFL is the most stable finding.Additionally, for ADHD, thinning of the macular thickness seems to be another informative parameter, in contrast to measurements of ganglion cell-associated layers, which were mostly non-significant (Atas et al., 2020;Li et al., 2021;Sánchez-Guillén et al., 2020).Nevertheless, Tünel and Kes ¸kek (2021) found a significant reduction in GCC in patients with ADHD.Interestingly, this was the only study with adult participants suggesting that apoptosis of RGCs and thus thinning of ganglion cell-associated parameters might occur later in the disease progression.Retinal thinning in conditions with DA overactivity might be a consequence of enhanced DA oxidation, which damages essential cellular (i.e., mitochondrial) functioning and further results in cell death (Luo and Roth, 2000;Xu and Yang, 2022).This would be in line with the inverted U-shaped function of DA on cognition, whereby too low and too high DA levels result in cognitive dysfunctioning, but might also cause retinal thinning.Further, when interpreting the thinning of the pRNFL from research on patients with schizophrenia, it is important to bear in mind that the aforementioned meta-analyses were not able to assess the effect of antipsychotic medication on retinal parameters (Kazakos and Karageorgiou, 2020;Lizano et al., 2020;Pan et al., 2018).As it has been suggested that antipsychotic medications (i.e., DA antagonists) may have toxic effects on the retina (Brandies and Yehuda, 2008;Silverstein et al., 2020;Silverstein and Rosen, 2015), care must be taken when interpreting these results as they may be confounded by medication use.
With regard to ASD, there are some inconsistencies in the pattern of  * showed a negative correlation with age the results.The findings of Gialloreti et al. (2014) are in line with the research on other diseases characterized by dopaminergic dysregulation.In contrast, García-Medina and colleagues (2017) reported increased layer thicknesses in patients with ASD and no significant differences in retinal parameters in another study between patients with ASD and neurotypical participants (Garcia-Medina et al., 2020).While the dysregulation of dopaminergic pathways in ASD is widely recognized, no consensus on the etiology has yet been found.While some studies suggest that autistic behaviors arise from a general DA hypoactivity, others reported positive effects of DA antagonists (i.e., risperidone) regarding stereotypic and social behavior (for a review, see Pavȃl, 2017).Further explanations for conflicting results could be co-existing conditions (e.g.hyperactivity or obsessive-compulsive behaviors) and associated medication administration that was not reported/evaluated.In addition, methodological differences such as sample size and age range of participants could have influenced the results.
Focusing on AD, several recent meta-analyses show a decline in retinal thickness across several parameters like RNFL, GCL-IPL, or macular volume and thickness (Chan et al., 2019;Coppola et al., 2015;den Haan et al., 2017;He et al., 2012;Sheriff et al., 2023;Thomson et al., 2015).The thinning of the RNFL was the most robust finding across the meta-analyses.However, it could also represent a selection bias of retinal parameters, given not all analyses took macular and ganglion-cell associated parameters into account.
Most evidence supports reduced retinal thickness in disorders associated with dopaminergic dysregulation and in older adults (who usually show a decline in DA functioning); particularly robust findings involve pRNFL, GCL and GCC.Unfortunately, there are still inaccuracies and discrepancies in the terminology of the retinal parameters studied.While some authors specified the location of the parameters studied (e.g.measurement of pRNFL or mRNFL), others did not.In addition, some studies examined only RNFL or GCL, whereas others examined all retinal layers.Hence, we cannot exclude the possibility that other parameters may be more suitable than pRNFL, CGL, or GCC for determining dopaminergic function.Therefore, it is imperative that future studies are more methodologically thorough and that results are reported using consistent terminology.Further, findings of retinal thinning in the elderly must be interpreted cautiously regarding its association with DA dysfunctions, given that other age-related conditions like brain atrophy (Shi et al., 2019) and comorbidities like diabetes and hypertension might also influence retinal thickness (Di Marco et al., 2022;Leley et al., 2021;Liu et al., 2020;Porta et al., 2005).However, this evidence does not speak against the hypothesis as DA being an important factor in retinal functioning as it is suggested that DA signaling might be altered in diabetes (Fiory et al., 2019), can affect blood vessels, and influences heart rate and blood pressure (Banday and Lokhandwala, 2008;Neumann et al., 2023).Moreover, there is a wealth of evidence connecting DA abnormalities with factors often observed in aging like cognitive decline (Bäckman et al., 2010(Bäckman et al., , 2006)), inflammation (Hunter et al., 2007;Lohr et al., 2017), and oxidative stress (Luo and Roth, 2000), which all have been shown to correlate with retinal thickness (Jadeja and Martin, 2021;Ko et al., 2018).In sum, it is likely that age-related retinal thinning might emerge from several (interacting) factors, whereby DA imbalances might only be only one of them.
In sum, retinal thinning is an omnipresent condition in several neurodegenerative diseases, but also in non-pathological aging.We suggest that DA abnormalities, which are evident in all these conditions, might be one crucial underlying mechanism of retinal thinning.However, future research is needed to illuminate the interplay with potential confounding variables.

OCT as a potential surrogate marker of cognition?
The studies reviewed so far pointed out that pharmacological manipulations of DA can influence OCT parameters and that patients suffering from dopaminergic abnormalities display atypical OCT parameters.This evidence suggests that OCT can be regarded as a surrogate marker to assess dopaminergic function.Given that DA plays a crucial role in regulating memory, attention, and executive functions (Aquili, 2020;Arnsten et al., 2015;Logue and Gould, 2014;McNab et al., 2009;Ott and Nieder, 2019;Weber et al., 2022), it is particularly important to consider that a natural decline in DA levels in the brain (Karrer et al., 2017) and alterations in the striato-midbrain-striatal loop (Rieckmann et al., 2011) are observed during aging, which can contribute to various cognitive impairments seen in severe cases of age-related neurodegeneration like AD (Krashia et al., 2019) as well as in normal aging (Bäckman et al., 2000;Erixon-Lindroth et al., 2005;Floel et al., 2008;Murman, 2015).To date, several studies investigated the modulating role of DA in (working) memory (Cools et al., 2008, p. 200;Cools and D'Esposito, 2011;Kourosh-Arami et al., 2023;Papenberg et al., 2020;Weber et al., 2022).By modulating the synaptic plasticity (Broussard et al., 2016;Espadas et al., 2021;McNamara et al., 2014;Rosen et al., 2015;Yamasaki and Takeuchi, 2017), DA facilitates not only memory formation, retrieval, and consolidation (Dalley et al., 2005;de Lima et al., 2011;Kramar et al., 2021;Rieckmann et al., 2011) but also maintaining and updating task-relevant information (Cools and D'Esposito, 2011;Durstewitz et al., 2000;Durstewitz and Seamans, 2008;Frank et al., 2007;Ott and Nieder, 2019).As DA decreases, older adults may experience difficulties in forming new memories or retrieving existing ones.This can manifest as forgetfulness, the inability to recall information quickly or to actively update information under time pressure.Further, DA increases the firing rate of PFC neurons, thereby enhancing the selection and processing of relevant information while filtering out distractions (Li et al., 2001;Servan-Schreiber et al., 1990;Warren et al., 2016).Due to the DA levels decrease, old people may experience difficulties in staying attentive and may be more easily distracted, affecting their ability to concentrate on tasks.However, results regarding age-related deficits in attention are heterogeneous.Recently, (Veríssimo et al., 2022) reported an overall decrease in the alerting network in older ages.Regarding sustained attention, meta-analytical evidence suggests that older adults use a more prudent strategy when allocating attentional resources resulting in slower reaction times but also smaller error rates (Vallesi et al., 2021), whereas no aging effects were found for selective attention (Verhaeghen and Cerella, 2002).Moreover, deficits in executive functions such as inhibition, set-shifting, and decision-making, can also be affected by decreasing DA levels in the PFC (Logue and Gould, 2014;Ott and Nieder, 2019).With dopamine decline in aging, older individuals may show alterations in executive functioning (Berry et al., 2019;Floresco and Magyar, 2006;Lighthall, 2020;Verhaeghen et al., 2003;Verhaeghen and Cerella, 2002;Veríssimo et al., 2022) and find it more challenging to switch between different approaches (Wasylyshyn et al., 2011), or considering alternative solutions or may experience difficulties in generating and executing complex plans, anticipating future outcomes, and effectively solving novel or challenging problems.
Hence, given that DA is a neurotransmitter, which is directly involved in regulating cognitive processes, we speculate that OCT may be a valuable surrogate marker for the assessment DA modulated cognitive performance, particularly useful in old age.To support this idea, we will give an overview of findings on the relationship between cognitive functioning and OCT parameters in patients suffering from dopaminergic abnormalities.There are few studies investigating this question in patients with neurodegenerative diseases (e.g., PD, MS, and AD).In addition, we looked into the findings of studies that investigated this relationship in healthy older adults, as this non-pathological population also has reduced dopaminergic function (Karrer et al., 2017).All studies outlined below are summarized in Table 2.
In MS, most studies suggest an association between retinal thinning and cognitive performance (i.e., information processing speed) for pRNFL and ganglion cell-associated layers (Abdel Naseer et al., 2019;Birkeldh et al., 2017;Bsteh et al., 2019), while one study failed to replicate this relationship (Frau et al., 2018).Regarding PD, to our knowledge, there is, only one study that has investigated the relationship between cognition and retinal layers.Elkhatib et al. (2019) reported a positive correlation between the superior quadrant of pRNFL and global cognition (measured with the Montreal Cognitive Assessment; MoCA).In line with MS studies, the finding of this study in PD patients also suggests that retinal thinning is associated with poor cognition.Regarding AD, findings are heterogenous.While some studies reported a significant relationship between the performance in the Mini Mental State Exam (MMSE), a common screening test in AD, and OCT parameters (Ascaso et al., 2014;Cunha et al., 2016), the majority did not (Cipollini et al., 2020;Gao et al., 2015;Kesler et al., 2011;Lian et al., 2021;Polo et al., 2014).However, (Lian et al., 2021) found worse cognitive performance in several stand-alone tests to be associated with abnormal RNFL thickness, and another study reported a significant relationship between constructional abilities and macula volume (Cipollini et al., 2020).
Unfortunately, the studies that investigated the relationship between cognitive performance and OCT parameters in non-demented older adults used different paradigms to measure cognitive functioning.Therefore, the findings are quite heterogeneous and hard to compare.While two studies found a linear association between different memory domains (i.e., logical and delayed memory, respectively) with pRNFL (Shi et al., 2019) and CGL-IPL (Liu et al., 2019), one study did not find such a relationship for immediate or delayed memory (Shi et al., 2020).Notably, while Méndez-Gómez et al. (2017) did not find a significant correlation between pRNFL and cognitive measures in a longitudinal study at baseline, the data revealed an association between pRNFL and future decline in delayed memory (i.e.thinner pRNFL was associated with increased decline).Interestingly, Liu et al. (2019) suggest a non-linear relationship between GCL-IPL and global cognition in 605 community-dwelling elders, whereby increased or decreased layer thickness was associated with deficits in cognition.This finding is particularly intriguing because it suggests that the relationship between OCT parameters and cognition might follow an inverted U-shaped function, where increased and decreased layer thickness are associated with poor cognitive function.The authors argue that a thickening of GCL-IPL could be a consequence of increased amyloid beta deposition and neuroinflammatory activity.This is in line with findings reporting elevated amyloid beta deposition and the up-regulation of inflammatory responses (i.e., "inflammaging") in normal aging (Lim et al., 2013;Onyango et al., 2021;Rodrigue et al., 2009;Tangestani Fard and Stough, 2019).
In conclusion, taking into account the natural process of DA decline in aging, our observations support the connection between cognitive functioning and OCT parameters in individuals dealing with dopaminergic abnormalities.Despite the need for additional research, the reviewed findings support the idea of OCT as potential surrogate marker for DA-controlled cognitive activities.This might be particularly relevant as individuals age and are at an increased risk for cognitive decline.The investigated studies suggest that thinner retinal parameters were associated with poor cognitive function.Consequently, as people advance in age, OCT could help to manage and monitor cognitive health in aging.

Current challenges and future directions of OCT as a potential surrogate marker of dopaminergic neuromodulation and cognition in ageing
This article presents the idea of OCT as a potential surrogate marker to assess dopaminergic neuromodulation and its related cognition in ageing.Here, we summarize the most important conclusions that can be drawn from a field that is still in its infancy and delineate concepts for future research.Nevertheless, the reviewed literature suggests that OCT is a valid surrogate marker of dopaminergic function, as elderly populations with dopaminergic abnormalities (e.g., people with PD, AD, and not to forget non-demented older adults) show a reduction in several OCT parameters (see Table 1).Notably, the decline dopaminergic efficiency in old age can be attributed to several factors.Meta-analytical evidence, thereby, suggests a negative effect of age on DA transporter and receptor density but no effect on DA release in ageing (Karrer et al., 2017).However, the loss of DA receptors vary across brain areas in ageing (Malén et al., 2022;Seaman et al., 2019) and their functional modifications can include a decrease in the number of receptors, alterations in their signaling pathways, or changes in their affinity for DA, which may result in reduced responsiveness to DA stimulation.Further, there is evidence for alterations in DA syntheses due to the gradual loss of DA-producing neurons in ageing (Gasiorowska et al., 2021) and to the loss of the enzyme tyrosine hydroxylase in the striatum in ageing (Salvatore et al., 2019).With respect to TH, in old age, the activity of this enzyme and other enzymes involved in DA synthesis may decline, leading to reduced conversion of L-tyrosine to L-DOPA and subsequently to DA.However, a recent study by (Ciampa et al., 2022) showed an increase of striatal DA synthesis capacity (measured by the uptake rate of the PET tracer [18 F]Fluoro-l-m-tyrosine) in the context of cortical thickness reductions (especially in posterior parietal areas) in the elderly.The authors conclude that the upregulation of DA synthesis capacity hereby might represent a mechanism of cognitive resilience in older adults as cognitive performance was preserved even in states of brain atrophy.Besides syntheses and transmission of DA, alterations in vesicular monoamine transporter (VMAT) functioning and availability might also cause disruption of DA functioning in aging (Bohnen et al., 2006;Frey et al., 1996;Ota et al., 2023).Given that VMAT is responsible for packaging DA into vesicles for storage and release, changes in the expression or function of VMAT may lead to impaired dopamine packaging and reduced availability for release in older adults.However, alterations in DA release during ageing are still discussed (Gasiorowska et al., 2021).
Additionally, based on the evidence that DA is directly involved in regulating cognitive processes and that DA declines in old age (see Section 4), the reviewed studies showed a significant association between retinal parameters and cognition (see Table 2).Therefore, it is all the more critical to outline directions for future research.We will propose theory-driven ideas why OCT may be an appropriate trait marker for the underlying mechanisms between dopaminergic and cognitive functioning.We will discuss the role of DA in regard to the concept of gain control, two antagonistic dopaminergic pathways that affect cognitive control and how this fits into the framework of metacontrol.Moreover, we will relate the existing assumptions to previous research findings with OCT.Finally, we will propose OCT parameters as a biomarker for individualized "ageing medicine" to optimize cognition in response to DA based pharmacological interventions.

In search of the mechanism underlying the cognitive effects linked to retinal thinning
The studies reviewed in this article indicate that OCT is a valid

Table 2
Overview of studies correlating OCT parameters with cognition in populations suffering from DA dysregulations, such as non-pathological aging, MS and PD.For statistically significant effects, the pattern of results resulting from the association OCT parameters and cognition is described under "Cognitive findings".

Retinal and cognitive findings Study
Non surrogate marker of dopaminergic functioning.We suggest that changes in retinal thinning and DA signaling likely translate into changes in "cognitive control", which regards cognitive processes that are necessary for goal-directed thought and action, like working memory (WM) and inhibition, as well as mental flexibility (Diamond, 2013).Aging is associated with a decline in cognitive control abilities, which can manifest as decreased performance in tasks requiring complex problem-solving, attentional shifting, and inhibition of irrelevant information (Paxton et al., 2007).Interestingly, alterations in neurotransmitter systems, such as reduced DA availability, have been implicated in the decline of cognitive control in older adults.Thereby, a correlative triad between aging dopamine and cognition has been illuminated by several studies using PET and SPECT scans showing that age-related decreases in DA function correlate with deficits in both cognitive and motor functioning (Bäckman et al., 2006;Braver and Barch, 2002).Furthermore, it is hypothesized that older adults show less dynamic neurotransmitter response and higher performance variability during demanding cognitive activities compared to (younger) adults due to an age-related loss of DA receptors and a reduction in the capacity for dopaminergic modulation during cognitive challenges (Bäckman et al., 2010;Li et al., 2010;MacDonald et al., 2012).When looking at OCT parameters and DA-driven cognitive functions, understanding the neurophysiological role of DA and its translation to cognition is essential to allow theory-driven predictions regarding their relationship.As mentioned previously, DA affects different dopaminergic receptor types (i.e., D1 and D2 receptor families), which are distributed unequally in the brain and in the retina (Beaulieu and Gainetdinov, 2011;Camps et al., 1990;Roy and Field, 2019).Moreover, pRNFL and ganglion cell associated retinal layer thicknesses correlate with cognition in populations displaying dopaminergic abnormalities (see Table 2).Taking into consideration the similarities in the unequal distribution in terms of D1 and D2 receptors in the brain and in the retina, we suggest that retinal OCT parameters may mainly mirror DA functioning in the PFC rather than DA functioning in the striatum.Accordingly, given that the focus of this section will be on understanding the mechanism underlying the cognitive effects linked to retinal thinning, we will mainly address cognitive functions mediated by the PFC in relation to ageing.

Retinal thinning and cognitive control
According to the framework of metacontrol (Beste et al., 2018b;Goschke, 2000;Hommel and Colzato, 2017), cognitive control emerges from the interplay of two dopaminergic pathways: the mesofrontal pathway (originating in the ventral tegmental area and targeting the PFC) and the nigrostriatal pathway (originating in the substantia nigra and targeting the striatum).Notably, Cools and D'Esposito (2011) suggested that cognitive control is a multifaceted phenomenon arising from the balance between cognitive stability (implicating the PFC) and cognitive flexibility (involving the striatum).As such, affecting the levels of DA will have paradoxical effects on these two distinct components, depending on which brain area is affected.Focusing on cognitive functions mediated by the PFC, it has been suggested (Durstewitz and Seamans, 2008) that two different representational states can be prominent depending on the distinct distribution of DA receptors in the PFC.Specifically, the authors propose that D1 and D2 receptors in the PFC promote either a "closed" or an "open" processing state, respectively.Thereby, a D1-driven "closed" state facilitates cognitive maintenance/persistence, while a D2-driven "open" state facilitates mental flexibility by updating (gating) of cognitive representations.Reconsidering the unequal distribution of D1 and D2 receptor types in the CNS, one could, therefore, assume that higher activity of the ventral tegmental area might activate more D1 receptors in the PFC along the meso-frontal pathway (i.e., driving cognitive control towards maintenance/persistence), while a stronger activation of the substantia nigra would activate more D2 receptors in the striatum along the nigrostriatal pathway (i.e., driving cognitive control towards flexibility).Taken together, the assumption of an interplay of the mesofrontal and the nigrostriatal pathway in the brain shaping cognitive control (Cools and D'Esposito, 2011) and the modeling-based approach of (Durstewitz and Seamans, 2008) point to two similar frameworks that might possibly even be considered functionally equivalent.The existing body of research (for a review, see Hommel and Colzato, 2017) suggests that cognitive control is not a unitary function but rather emerges from the interplay of probably two antagonistic pathways or processing modes (i.e., between the mesofrontal vs. nigrostriatal pathway or cognitive persistence vs. flexibility, respectively) (Kehagia et al., 2010;Stokes, 2015).Like any other antagonistic system, cognitive control modes exclude each other.For example, the more one shifts their attention to task-relevant information and, therefore, ignores task-irrelevant information, the more one is likely to oversee other rewarding or suitable opportunities of action.Because none of these two extremes will generate a satisfactory outcome, according to the framework of metacontrol (Hommel, 2015;Hommel and Colzato, 2017), the balance between persistence and flexibility is crucial.Within the framework of metacontrol, given that the retina is rich in D1 (Roy and Field, 2019), we would assume that retinal thinning might predict cognitive persistence relying on the PFC along the mesofrontal pathway, but not cognitive flexibility relying on the striatum.With respect to the three facets of cognitive control (i.e., flexibility, inhibition, and WM) (Diamond, 2013) and on the basis of the previous assumption that RNFL and ganglion-associated retinal layers may be surrogate markers of dopaminergic activity, we hypothesize OCT parameters to be most predictive for tasks addressing inhibition and action control rather than flexibilityparticularly in older individuals (Wasylyshyn et al., 2011) or diseases characterized by dopaminergic alterations like PD (Manza et al., 2017) as the former functions depend on D1-modulated DA activity (Beste et al., 2016;Stock et al., 2014), which is likely to be impaired in the elderly (Karrer et al., 2017).When considering WM functions, converging evidence suggests the maintenance of task relevant information to depend on PFC function (Durstewitz et al., 2000), whereas updating and interruption of action goals are mediated by the striatum (Cools and D'Esposito, 2011;Frank et al., 2007).Again, especially in aging, OCT parameters might be more sensitive to the maintenance than the updating processes of WM representations given that greater losses of D1 than D2 receptors were reported in older adults (Karrer et al., 2017).
Taken together, retinal thinning may predict cognitive persistence via the D1-mediated mesofrontal pathway.Due to the predominance of D1 receptors in the retina, OCT parameters are suggested to be more indicative of inhibition and action control tasks, highlighting the potential utility of retinal measures in predicting cognitive functions.

Retinal thinning and basic neural information processing principles relevant to cognition across the life span
Another crucialif not the most importantfactor when looking at cognitive control processes, and especially so in the context of ageing (Li and Rieckmann, 2014;Li and Sikström, 2002;Schmitz et al., 2014), is the ability to distinguish between task-relevant signal and task-irrelevant noise during information processing (Adelhöfer et al., 2018;Pertermann et al., 2019aPertermann et al., , 2019b;;Salinas and Thier, 2000;Servan-Schreiber et al., 1990).For example, a high signal-to-noise ratio (SNR), reflects a better distinction between relevant and irrelevant information and thus shifts cognitive control to a state of maintenance/persistence (Aston-Jones and Cohen, 2005;Bensmann et al., 2019Bensmann et al., , 2018)); therefore, a low SNR (i.e., more noise) shifts cognitive control processes towards a state of updating/flexibility (Gureckis and Love, 2009).This SNR modulation or improvement, respectively, is mirrored by an increase of gain control (Li and Rieckmann, 2014;Servan-Schreiber et al., 1990;Yousif et al., 2016;Ziegler et al., 2016), which is an overarching principle in the CNS found at sensory, cognitive (Adelhöfer et al., 2018;Bensmann et al., 2018;Salinas and Thier, 2000;Servan-Schreiber et al., 1990), and motor processing levels (Greenhouse et al., 2015;Thura and Cisek, 2016).Interestingly, DA and other catecholamines modulate the SNR of neuronal activity (Aston-Jones and Cohen, 2005;Cohen et al., 2002;Kroener et al., 2009;Li et al., 2001;Li and Rieckmann, 2014;Nieuwenhuis et al., 2005;Rolls et al., 2008;Servan-Schreiber et al., 1990;Vander Weele et al., 2018;Winterer and Weinberger, 2004;Yousif et al., 2016;Ziegler et al., 2016) by sharpening the distinction between irrelevant neural noise and relevant neural input, therefore modulating gain control processes (Li et al., 2001;Servan-Schreiber et al., 1990;Warren et al., 2016).This improvement can be illustrated as an increasing slope of the sigmoid input-output function (Fellous et al., 2003;Li et al., 2001;Salinas and Thier, 2000;Servan-Schreiber et al., 1990) (Fig. 2).Thus, high gain control (i.e., the steep slope of the sigmoidal function, orange curve) would be associated with higher cognitive functioning (Greenhouse et al., 2015;Salinas and Thier, 2000;Thura and Cisek, 2016).During the aging process, the decline of DA (Bäckman et al., 2010(Bäckman et al., , 2006) makes it more difficult for neurons to differentiate between relevant input and irrelevant neuronal noise (Berry et al., 2016;Li et al., 2010;Li and Rieckmann, 2014).This results in a decrease in neuronal gain control which negatively affects cognition in older individuals noise (Berry et al., 2016;Gazzaley et al., 2005;Li et al., 2010;Li and Rieckmann, 2014).On the contrary, optimal DA levels in healthy young individuals have been found to enhance the SNR and gain control, promoting cognitive processes (Adelhöfer et al., 2018;Bensmann et al., 2019Bensmann et al., , 2018;;Beste et al., 2018a).That is, the reduction in DA levels that occurs with aging results in decreased or less effective gain control in comparison to younger adults, where the slope of the sigmoid function is more pronounced (Passow et al., 2017).Accordingly, given that retinal thinning seems to be a reliable surrogate marker to assess dopaminergic functioning, we expect people with pathological low OCT parameters (see Fig. 2; flatter slope of the sigmoidal function, yellow curve), such as older adults, to be associated with lower/less efficient gain control, as compared to people with average/high OCT parameters (orange curve), such as younger adults.Moreover, gain control is assumed to be most important in high demanding tasks where cognitive control processes are highly engaged, with the present evidence suggesting that gain control mechanisms are modulated by the catecholaminergic system (Adelhöfer et al., 2018;Bensmann et al., 2019;Pertermann et al., 2019a).Hence, resuming the previous literature, the SNR (as a surrogate marker for gain control) might be the most plausible neural candidate underlying the metacontrol framework of cognitive control.
Given that retinal thinning is linked to decreased DA levels affecting D1 receptor families, we suggest that a substantial amount of retinal thinning's effects onto metacontrol might produce a decrease in gain control in cognitive persistence.Specifically, we expect a correlation between retinal thinning and cognitive persistence (i.e., more thinning would predict more deficits when facing potential distractors) but not cognitive flexibility because of the D1 predominance in the retina.This correlation might be further pronounced in the elderly, as age has a greater negative effect on D1 than D2 receptors (Karrer et al., 2017) and, crucially, D1 receptors are important for the stabilization of neural representations by narrowing spatial tuning (Arnsten, 2011).Moreover, we would expect this correlation to follow an inverted U-shaped function, as this was found for the relationship between stimulation of D1 receptors and PFC functions in animals (Cools and D'Esposito, 2011;Goldman-Rakic et al., 2000;Williams and Castner, 2006) and humans (Takahashi et al., 2008) before.To test our hypothesis it would be ideal to use two task-switching paradigms independently developed by Dreisbach et al. (2005) and by Armbruster et al. (2012).These paradigms are suited because they are uniquely able to assess, at the same time, cognitive flexibility (as indicated by flexible switching between task rules) and cognitive persistence (performance in the presence of irrelevant distractors).Interestingly, meta-analytical evidence suggests a negative impact on global switch costs, which are associated with the maintenance of task information in WM, but not in local switch costs, which are closer related to the updating of relevant task information (Wasylyshyn et al., 2011).
To summarize, gain control and SNR, which are influenced by DA and other catecholamines, play a pivotal role in cognitive processes, particularly in aging.The relationship between DA, its receptor subtypes, and cognitive functions allows us to hypothesize how OCT parameters might be used as surrogate markers for DA-driven cognition and behaviors, especially those relying on the D1 receptor system in ageing.In the next section, based on the evidence that dopaminergic functioning and some facets of cognitive processes are related in a nonlinear way (Cools et al., 2019;Cools and D'Esposito, 2011;Ott and Nieder, 2019;Takahashi et al., 2012Takahashi et al., , 2008)), we will discuss the idea of employing retinal thinning as a trait marker to optimize cognition with respect to tailored pharmacological interventions for DA-related change in ageing and age-related disorders.

OCT parameters as a biomarker for individualized medicine to optimize cognition in response to DA based pharmacological interventions across the life span
Cognitive performance and DA levels are unlikely to be associated linearly.Indeed, it is assumed that cognitive functioning and DA levels are related in an inverted U-shaped function; with best performances being related to medium DA levels and deficits in performance with very high or very low levels, respectively (see Fig. 3A).Thereby, this overarching pattern could not only be observed in (young) adults (Colzato et al., 2021;Cools and D'Esposito, 2011) but also in populations with age-related diseases like PD (Cools, 2006).Along the same lines, Liu et al. (2019) found GCL-IPL and global cognition associated with an inverted-U-shape function: individuals with medium GCL-IPL thickness outperformed individuals with low or high layer thickness.Two implications for pharmacological interventions arise from these observations.First, whether the administration of a dopaminergic agonist enhances or impairs cognitive performance depends on the individual's baseline level of DA availability.Therefore, it is more likely to impair one's performance by drug administration if the person's baseline level is close to its theoretical optimum.Hence, the use of DA markers to index baseline DA levels would be extremely advantageous to optimize cognition in response to DA based pharmacological interventions.Indeed, this personalized approach can lead to more effective and targeted treatments for age-related cognitive decline, ultimately improving the overall quality of life for older adults.However, in the past, a failure to properly understand and use surrogate markers has hindered their broad adoption and translation to individualize and optimize clinical and cognitive outcomes (Burke, 2016).Here, we propose that only a mechanistic approach making use of theory-driven markers to locate individuals on the inverted U-shaped function between DA levels and cognitive functioning can guide us to successfully develop Fig. 2. Schematic illustration of potential gain modulation processes in people with low and high OCT parameters likely corresponding to low and high DA baseline levels.With the same input value (e.g., cognitive effort), individuals with average OCT parameters would therefore achieve a better cognitive performance.
individualized medicine interventions.Consistent with this idea, the second implication entails that whether a given DA agent produces benefits or impairments depends on its dose and the individual "status quo" that can, for example, rely on the individuals' baseline performance level, see Fig. 3B.We suggest that performance in cognitive tasks driven by DA will be predicted by individual differences in OCT parameters (our surrogate marker of DA levels) following an inverted-U-shape function: people with intermediate OCT parameters will outperform people with low or high OCT parameters.For instance, given that OCT parameters have been found to correlate with the severity of PD (Ahn et al., 2018), the same OCT markers could be used to locate different patients, or the same patient over time, on the inverted U-shaped function to determine an individualized dose to achieve a successful pharmacological treatment.That is, the thinner the retinal parameters of the PD patients will be, the higher the levodopa dose is likely to be required by the patient.This consideration is important for two reasons.First, studies of PD in humans and rodents showed apoptosis of retinal dopaminergic neurons (Ahn et al., 2018;Lee et al., 2020;Tatton et al., 1990) even before a dopaminergic loss in the brain (Normando et al., 2016;Prudêncio et al., 2002).This means that OCT parameters might be particularly time-sensitive markers to rely on to determine a successful and individualized DA based pharmacological intervention in PD.Second, in PD literature, a well-known phenomenon is dopaminergic "overdosing", which leads to impulse control deficits and impairments in decision-making (Kehagia et al., 2012).Hence, also in this case, OCT parameters might be useful to optimize the dose in order to limit the dopaminergic "overdosing" issue in PD.
Three lines of research support this rationale described above.First, several studies in non-pathological elderly and in individuals suffering from DA abnormalities found cognitive performance related to individual differences in OCT parameters (see Table 2).Second, interindividual baseline levels of DA are crucial for cognition-enhancing drugs (e. g., amphetaminergic agents like Adderall® or Dextroamphetamine or Levodopa).Effects of amphetaminergic drugs also follow an inverted Ushaped function, in which low-performing individuals benefit from use, whereas cognitive functioning is impaired in high-performing individuals (Farah et al., 2009;Mattay et al., 2000).Specifically related to aging, several studies have shown that low to moderate doses of levodopa can have beneficial effects on specific cognitive domains, such as attention, WM, and executive function, in individuals with PD and age-related cognitive decline (Cools, 2006).These cognitive improvements are thought to be due to the normalization of DA levels in the substantia nigra.However, at higher doses, levodopa can lead to dopaminergic overstimulation, which may have negative cognitive effects.High doses of levodopa have been associated with impulsivity, attentional deficits, and impaired executive functions (Cools, 2006).Moreover, evidence from pharmacological studies on agents targeting the DA system leads to the assumption that contrasting effects of dopaminergic drugs in humans (Cools et al., 2008;Urban and Gao, 2014) may also be a consequence of DAs non-linear relationship with cognitive performance (see Fig. 3).Studies in animals have corroborated this idea and suggest that these contrasting effects mirror interindividual baseline DA levels (Arnsten, 1998;Phillips et al., 2004;Williams and Goldman-Rakic, 1995;Zahrt et al., 1997).The authors assume that low baseline DA levels correspond to poor WM functioning, while high baseline levels of DA result in good WM functioning.Further support comes from PET studies in humans showing a relationship between cognitive performance and DA receptor binding, supporting the idea that, indeed, lower D1 and D2 receptor availability in PFC is associated with deficits in WM across the lifespan (Aalto et al., 2005;Bäckman et al., 2000;Landau et al., 2009;Nyberg et al., 2016;Takahashi, 2013;Takahashi et al., 2012).Next, these findings are consistent with results from psychopharmacological studies in healthy participants investigating the effects of dopaminergic drugs on cognition which are known to depend on baseline levels of WM capacity (Daneman and Carpenter, 1980;Salthouse and Babcock, 1991) or prior learning experiences (Mückschel et al., 2020a(Mückschel et al., , 2020b)).Along the same lines, dopaminergic agonists show positive effects on participants with low WM capacity and adverse effects on participants with high WM capacity (for a review, see Cools et al., 2008), especially in tasks regarding set-shifting (Cools et al., 2007;Frank and O'Reilly, 2006;Kimberg et al., 1997), updating (Frank and O'Reilly, 2006;Mehta et al., 2000), and WM retrieval (Gibbs and D'Esposito, 2005).Third, our line of reasoning is supported by previous findings from pharmacogenomics where homozygotes for the Val-allele and the Met-allele of the COMT val158met single nucleotide polymorphism are known to display a distinct cognitive profile following an inverted-U-shaped function in response to DA agents (Schacht, 2016).Indeed, a critical review examining 25 studies revealed that DA agents acting on increasing cortical DA levels enhanced cognitive processes the most in individuals homozygous for the Val-allele (i.e., individuals characterized by decreased cortical DA levels).In contrast, antipsychotic medications enhanced them the most in individuals homozygous for the met-allele (i.e., individuals characterized by increased cortical DA levels) (Schacht, 2016).Moreover, investigations of the association Accordingly, increasing DA level via DA agonists by the amount of Δ (depending on the dose received) will enhance cognitive performance in I, have little effect on II, and impair cognitive performance in III.Customized treatment for DA age-related changes can greatly benefit from this approach.Tailored medicine using OCT takes into consideration the changes in DA levels that occur with age, allowing for personalized interventions to target cognitive needs and challenges associated with aging.
between gene polymorphisms in aging and reductions in several aspects of dopamine function, including receptor availability, dopamine synthesis, and transporter density (Li et al., 2013(Li et al., , 2010;;Papenberg et al., 2014Papenberg et al., , 2013) ) suggest that these reductions correlate with decreased performance in a range of cognitive tasks, such as episodic memory, WM, and executive functioning.The research mentioned above highlights the importance of the dopamine system in maintaining cognitive health and suggests that the age-related decline in dopamine function might be a central mechanism behind the cognitive aging process.Within the psychopharmacological perspective, it is crucial to consider the interplay between the inverted U-shaped function and the SNR, particularly in relation to the process of aging.This is because neurophysiological changes associated with aging, such as alterations in DA levels, can influence the optimal dosage of psychotropic medications and their effectiveness.Indeed, age-related changes in DA levels can affect the balance between neural signal strength and background noise, potentially impacting the therapeutic response to psychopharmacological interventions in older adults.Keeping in mind these interactions can inform us in tailoring pharmacological treatments, considering the specific neurobiological changes that occur with aging, aiming to maximize therapeutic outcomes while minimizing potential adverse effects.Especially gain control principles are crucial for (a) the modulatory aspects of the DA system and (b) for cognitive control processes.As explained in the previous section, DA and other catecholamines modulate the SNR (Aston-Jones and Cohen, 2005;Li et al., 2001;Li and Rieckmann, 2014;Nieuwenhuis et al., 2005;Servan-Schreiber et al., 1990;Yousif et al., 2016;Ziegler et al., 2016) and may increase gain control (illustrated as a steeper slope of the sigmoid input-output function; see Fig. 2).In line with this idea, unmedicated patients with ADHD exhibited a flatter slope compared to healthy controls (Pertermann et al., 2019a).In contrast, after the intake of methylphenidate, the slope of patients with ADHD became steeper-showing that the medication reduced neural noise (Pertermann et al., 2019a).Along the same lines, increased catecholamine concentrations resulting from methylphenidate administration have been shown to increase gain control (Adelhöfer et al., 2018;Bensmann et al., 2019Bensmann et al., , 2018)).That is, medical effects targeting the DA system can be evaluated using the slope suggesting that the slope may also be a valuable and new neurophysiological outcome measure in pharmacotherapy (Münchau et al., 2021).Translated into the inverted U-shaped function logic presented in Fig. 3B, people with low OCT are associated with less efficient gain control and are likely located on the left side of the curve.In contrast, people with high OCT are linked to optimal gain control and are probably located at the apex of the curve.Therefore, following the logic of an inverted U-shaped dose-effect relationship between DA and cognitive performance, except for conditions causing swelling of retinal layers (e. g., glaucoma), the intake of DA agonist drugs might produce cognitive impairments in people with average/high OCT parameters (already associated with a steep slope), but cognitive enhancement in people with low OCT parameters (linked to a flat slope).The other way around, the intake of DA antagonist drugs might produce cognitive enhancement in people with average/high OCT parameters but cognitive impairments in people with low OCT parameters.Within the perspective of individualized medicine, we believe that the inclusion of theory-driven markers, such as OCT parameters, for a personalized approach will be a potentially effective tool to optimize cognition by promoting cognitive benefits and mitigating cognitive deficits in response to pharmacological intervention.Hence, OCT parameters might represent a valuable biomarker for identifying people who are more prone to respond positively (or negatively) to DA agents.This approach might be particularly relevant in terms of tailored treatment for age-related changes.Indeed, individualized medicine based on OCT can account for age-related DA changes, ensuring that DA interventions are customized to address specific cognitive needs and challenges that arise with aging.However, even though there are established thresholds for ophthalmological pathologies and several studies (with N > 200) showed an age-related decrease of retinal thickness and suggested annual/decadal decline rates (Feuer et al., 2011;Lee et al., 2012;Mwanza et al., 2011;Patel et al., 2016;Peng et al., 2017), to date there are no established normative values of OCT parameters associated with (normal) cognitive functioning.Furthermore, while it is possible to demonstrate that DA decline and retinal thinning coincide with aging (see Section 3.2 and Table 1), there is no evidence of a direct causal relationship between these observations, and retinal thinning in the elderly may instead be explained by other age-related mechanisms such as vascular conditions (Di Marco et al., 2022;Leley et al., 2021;Liu et al., 2020;Porta et al., 2005), which might not be reported due to their omnipresence in the elderly.We therefore advise future research to explicit report such comorbid factors.Still, due to DAs manifold influences on motoric and cognitive functioning across the lifespan (e.g., Bäckman et al., 2010;Klein et al., 2019;Korkki et al., 2023;Li et al., 2010) the relationship between cognitive and retinal decline (see Section 4 and Table 2), and the embryological similarities between the central nervous system and the retina (see Section 1), we believe OCT might be a promising tool to evaluate dopaminergic functioning, even outside of diseases characterized by DA abnormalities.
Overall, we emphasize the nonlinear relationship between cognitive performance and dopamine (DA) levels, as an inverted U-shaped function is observed across the lifespan.We propose the use of DA markers to personalize pharmacological interventions and optimize cognitive outcomes based on individual baseline DA levels.Considering the interplay between inverted U-shaped function and SNR, especially in old age, allows psychopharmacological treatments to be tailored.We emphasize the importance of gain control principles and focus in particular on OCT parameters as potential biomarkers for individualized medical interventions targeting age-related changes in DA levels and cognitive health.

Conclusion
Previous work showed that the retina offers several features as biomarkers (e.g., microvascular changes or amyloid pathologies) for early diagnosis in specific neurodegenerative diseases like AD (Alber et al., 2020;Cabrera DeBuc et al., 2017;Ge et al., 2021;Koronyo et al., 2017Koronyo et al., , 2012)).However, the present article introduced the idea of using OCT and its derived parameters (e.g., RNFL, GCL-IPL, etc.) as potential surrogate markers of general dopaminergic neuromodulation and related cognition.Although research is still in its initial stages and more systematic studies are needed, we propose GCL, IPL and INL as DA-sensitive OCT parameters and RNFL as an early marker of RGC degeneration.However, to find convergent evidence, future studies need to be more methodologically thorough and results reported using consistent terminology.
Although the studies reviewed suggest that OCT might reflect both D1 and D2 (drug-induced) activity, we hypothesize that OCT parameters are predominantly related to the D1 receptor system due to the dominance of D1 receptors in the retina (Roy and Field, 2019).Given that D1 receptors are also dominant in the PFC, but not in the striatum (Beaulieu and Gainetdinov, 2011;Camps et al., 1990), it is possible that OCT parameters are primarily related to cognitive functions modulated by the D1 receptor system of the PFC, probably in an inverted U-shaped fashion (Takahashi et al., 2012(Takahashi et al., , 2008)).Overall, retinal thinning may reflect lower DA activity and should be associated with a decrease in gain control and poorer cognitive function, especially those that rely on maintenance/persistence of cognitive abilities (see Graphical Abstract).
We argue for a stronger emphasis on mechanistic theories in guiding future research to examine the mechanism underlying the cognitive effects linked to retinal thinning.In this respect, the framework of metacontrol could be particularly useful.Indeed, keeping in mind similarities in the unequal distribution in terms of D1 and D2 receptors in the brain and the retina, this framework enables to predict of interindividual differences in cognitive functioning based on one's ability to modulate cognitive control processes in terms of shifting between cognitive persistence (i.e.D1-modulated) and flexibility (i.e.D2modulated).
Finally, we would like to recommend employing OCT parameters as biomarkers for individualized medicine to optimize cognition in response to DA based pharmacological interventions.That is, individual differences in OCT parameters might enable us to assess signs of an individual's DA baseline level.Given the non-linear relationship between DA functioning and cognition, identifying an individual's DA baseline level would support the identification of people who would benefit or be disadvantaged by dopaminergic drug administration.
As our society continues to age significantly in the coming years and decades, it becomes increasingly important to focus on maintaining cognitive function and promoting healthy aging.However, the standard methods used to investigate DA levels in relation to cognitive decline are complex and expensive.OCT can act as an "initial stage detection system", enabling us to identify individuals at a higher risk as early as possible.This early detection will allow for the implementation of personalized and timely treatment approaches that may be more effective than a generalized approach.In addition, by intervening at an earlier stage, we may potentially slow down the progression of cognitive decline in aging.So, OCT not only will help us identify at-risk individuals, but also to pave the way for more tailored and efficient treatments with the goal of a more efficient management of cognitive aging.

Fig. 1 .
Fig. 1. (A) Illustration of retinal cell types and layers.(B) A scan of the eye generated by OCT, including the most relevant parameters.ELM = External limiting membrane; BM = Bruch's membrane; ILM = Inner limiting membrane.

Fig. 3 .
Fig. 3. Illustration of OCT parameters as biomarkers for individualized medicine to optimize cognition in response to DA-based pharmacological interventions.(A) Cognitive performance tends to relate to neurotransmitter levels, such as DA, following an inverted-U-shaped function, in a way that intermediate DA levels are predictive of the best cognitive performance.(B) DA agonist drugs increase DA levels and enhance or impair cognitive performance in different people depending on their baseline DA level as reflected by OCT parameters.For instance, individual I is characterized by a low baseline DA level (indexed by low OCT parameters), II by an intermediate DA level (indexed by intermediate OCT parameters), and III by a high DA level (indexed by high OCT parameters).Accordingly, increasing DA level via DA agonists by the amount of Δ (depending on the dose received) will enhance cognitive performance in I, have little effect on II, and impair cognitive performance in III.Customized treatment for DA age-related changes can greatly benefit from this approach.Tailored medicine using OCT takes into consideration the changes in DA levels that occur with age, allowing for personalized interventions to target cognitive needs and challenges associated with aging.

Table 1
Overview of meta-analyses and single studies (not included in the metaanalyses) on OCT parameters in populations suffering from DA abnormalities.