Ophthalmic manifestations in children with tuberous sclerosis complex

To report on the ophthalmic findings in children with tuberous sclerosis complex (TSC) in southern Sweden, and to investigate the frequency of refractive errors, strabismus and cerebral visual impairment associated with this condition.

are not regarded as pathognomonic.They are usually not located in the central macula, and thus rarely affect the visual acuity.However, there have been reports of central hamartomas, and associated complications such as bleeding and oedema, resulting in visual impairment.Achromatic patches constitute another retinal sign, often localized in the midperiphery, and previously described in 39% of the patients (Rowley et al., 2001).These are considered a minor diagnostic criterion, and thus of relevance in the diagnostic work-up even though they do not cause visual impairment.
TSC also has several neuro-ophthalmological manifestations.These are not included in the diagnostic criteria, but can have a major impact on visual function.The most significant of these are optic nerve hamartomas, elevated intracranial pressure, cranial nerve palsies and cerebral visual impairment (CVI) (Toldo et al., 2019;Wan et al., 2019).Optic nerve hamartomas rarely affect visual acuity, but in some cases they can cause severe visual impairment.It is also important to distinguish them from papillary oedema as a sign of elevated intracranial pressure due to other intracranial tumours such as subependymal giant cell astrocytoma (SEGA).Other signs of elevated intracranial pressure are cranial nerve palsies and visual field defects.In CVI the damage is located retrochiasmally and can affect visual acuity, visual field function and higher-order visual functions, leading to problems in visual interpretation affecting school activities and everyday life (Dutton & Jacobson, 2001).During ophthalmologic examinations it is important to conduct neuro-ophthalmological tests, including a colour vision test, visual field examination, fundus photography, optical coherence tomography (OCT) scans to assess the retinal cell layers, and to inquire about problems associated with CVI.However, it is often difficult for TSC patients to participate in these examinations due to intellectual and neuropsychiatric disabilities, making such examinations and assessments more difficult.Another possible reason for a reduced visual function is retinal toxicity resulting from vigabatrin exposure, used for the medical treatment of early onset interictal activity and epileptic spasms in TSC patients (Maguire et al., 2010;Riikonen et al., 2015).The function of the retina can be measured objectively with electroretinography (ERG), often performed at tertiary centres and, when necessary, under general anaesthesia.This technique makes it possible to monitor the retinal function during treatment with vigabatrin.
Although TSC was described already during the 19th century (Henske et al., 2016), studies on the ophthalmic manifestations of the disease are generally scarce.In particular, there is a lack of knowledge concerning the risk of refractive errors and strabismus in these patients.We have therefore compiled ophthalmic findings in a paediatric TSC patient cohort in the southern region of Sweden.

| M AT ER I A L S A N D M ET HOD S
This was a retrospective cohort study including all paediatric patients diagnosed with TSC in southern Sweden, born between 1983 and 2020, who had undergone an ophthalmologic examination.Approval was obtained from the Swedish Ethical Review Authority (No. 2020-05201) and the study was carried out according to the Tenets of the Declaration of Helsinki.
Data were obtained from the patients' most recent medical records on visual acuity, refractive error, need for spectacles, strabismus, nystagmus, CVI and fundus appearance focusing on astrocytic hamartomas and achromatic patches.The 10 children in whom fundus photographs (RetCam 3; Natus Medical Inc., Pleasanton, CA, USA) and handheld OCT scans (Envisu C2300; Leica Microsystems, Wetzlar, Germany) had been taken under general anaesthesia were reviewed again for additional retinal findings.Another five had fundus photographs that were also possible to review, taken at various ophthalmology departments in the southern healthcare region with varying fundus cameras.
Visual acuity was reported as Snellen decimal acuity, or Snellen equivalent for children below 4 years of age using a preferential looking test (Kasper P.L. acuity & contrast sensitivity cards © Konstantin Moutakis, https:// www.synta vlor.se).It was presented as median (min-max) values for the better and worse seeing eyes as well as the deviation from that expected in a healthy child of the same age since their absolute visual acuity results could not be compared between subjects of varying age.The expected level of visual acuity from children 8 years or older was 1.0 Snellen decimal acuity.Expected levels for younger children were defined according to American Academy of Ophthalmology's recommendations (Saluja et al., 2023).Refractive error was measured with an autorefractor under cycloplegia or sciascopy if the child was unable to cooperate in the autorefractor.A few patients were old enough and with good intellectual capability to be able to perform a subjective refraction.Strabismus was defined as all types of manifest and intermittent strabismus.The number of patients who had undergone ERG examination due to treatment with vigabatrin was noted, as well as the ERG results.
It is difficult to compare ffERG responses for children as their retina is not fully developed at birth, and since we do not have age matched reference values.Therefore, we used the ffERG recording at the start of vigabatrin treatment for each patient as its own reference and the expectation was improved responses while growing up.The most stable variable to compare is the 30 Hz flicker response (Chiang et al., 2022).

| Neuropsychiatric and motor disorders
Information on intellectual disability, attention deficit hyperactivity disorder (ADHD) and autism were retrieved from medical records (diagnosed with age-standardized diagnostic tests according to age).Motor impairment, as diagnosed by physiotherapists, was also noted.

| Statistics
Results are reported as median and range as the data are not normally distributed, except for age at last ophthalmological visit where mean and standard deviation was reported instead.

| R E SU LT S
Ophthalmological records were available for 50 of the 52 TSC patients identified in the southern health care region of Sweden.At the last registered ophthalmological examination, the mean age was 12.4 years (SD 7.2).Monocular visual acuity was possible to perform in 39 individuals and the median value was 0.95 Snellen decimal acuity (min: 0.05, max: 1.3) in the better seeing eye and 0.8 Snellen decimal acuity (min: no perception, max: 1.3) in the worse seeing eye.The median did not deviate from that expected for their age (range no difference from expected to −0.95 from expected visual acuity) in their best eye but deviated from the value expected for their age by −0.2 Snellen decimal acuity (range no difference from expected to −1.0 from expected visual acuity) in their worse eye.Refractive errors were the most common reason for a reduced visual acuity (Table 1).
Data on refractive errors at the final ophthalmological examination were available for 41 patients, while another five had full visual acuity without correction, and were indirectly considered to be most likely free of significant refractive errors (Table 2).Refractive errors were common in these individuals and in total, 61% (28/46) of the patients had some form of refractive error.There was no information on the use of spectacles in two patients, but among the rest of the 48 patients, 40% (19/48) wore spectacles.
Forty-four of the 50 patients had been examined for strabismus, and 16% (7/44) of them had been diagnosed: five with exotropia and two with esotropia.Three of the exotropias were alternating but none of the esotropias.Data on the occurrence of nystagmus was available for 45 patients, and only two of these (4%) were diagnosed as having the condition.
Retinal astrocytic hamartomas had been diagnosed in 16 of the 47 fundus examined patients (34%), with bilateral findings in nine (56%) of these (Table 3).Only one of the tumours was found in the central macula, and hence rarely affected the patient's visual acuity.It is sometimes difficult to detect retinal astrocytic hamartomas, even when using digital imaging, under general anaesthesia.OCT with a handheld instrument can then be valuable (Figure 1).Among the children that had fundus photographs available for revision, 11 of 15 (73%) had retinal hamartomas.
Achromatic patches were detected in 34% of the patients (16/47), and in 10 of the 15 patients (67%) that had fundus photographs available for revision (Figure 2).None had angiofibromas in the eyelid.One patient had a growing opticus astrocytoma leading to enucleation of the eye.This occurred before oral mechanistic target of rapamycin (mTOR) inhibitors were available to treat aggressive astrocytomas.This patient also developed hydrocephalus and optic atrophy in the remaining eye due to SEGA growth.
Almost half of the patients, 46% (23/50), had been treated with vigabatrin for some period in their life.Ten of these had been examined with ffERG, with findings of a reduced response in two patients and a tendency in one patient, resulting in dose reduction or cessation of the treatment.A recovery was observed at subsequent follow-up examinations of the two patients when they had been out of medication for more than 12 months (Table 4).These two patients were both examined with the Espion system every time and with the patient's initial  recordings when vigabatrin treatment was initiated as a reference value.The third patient was examined with different ffERG systems over time making it difficult to evaluate the responses, Nevertheless, a tendency of reduced responses was observed.None of the 50 patients had undergone any kind of CVI examination.

| DI SC US SION
The catchment area of the study (the southern counties in Sweden; Skåne, Kronoberg and Blekinge) has around 19 000 births annually, which during the study period would have yielded between 40 and 60 TSC patients given the international incidence numbers.We identified 52 individuals, in line with these numbers, and believe to have a near complete coverage of our cohort.Several well-known ophthalmic manifestations are associated with TSC.In the cohort studied here, 34% had been diagnosed with retinal astrocytic hamartomas and 34% with achromatic patches.The frequency of retinal astrocytic hamartomas was somewhat lower in this retrospective study on children than the 44%-50% described in earlier reports on mostly adult patients (Robertson, 1991;Rowley et al., 2001).The percentage of children demonstrating achromatic patches was also slightly lower than the 39% reported earlier (Rowley et al., 2001), most likely due to the lower general awareness among ophthalmologists of these retinal findings, but also because the peripheral retina can be more difficult to examine in children with intellectual disabilities and autism.Retinal changes may also be more frequent among adult patients.New technical abilities with both fundus photography and OCT vastly improve the detection rate, and this could affect our results since they were not available for the first couple of decades of the study period.
A considerable proportion of the patients had substantial refractive errors and required spectacles.There have been few publications on refractive errors in TSC patients, making comparisons with previous results difficult due to the variation in age of the study subjects, incomplete cycloplegia in previous studies and different  limits on the degree of refractive errors reported (Rowley et al., 2001).The 61% with refractive errors in our study cohort indicates that it is a substantially more common problem than reported in healthy 10-year-old children, where it has been reported that 3.6% have hyperopia of ≥+2D, 7.8% myopia ≤−0.5D and 4.1% astigmatism ≥1D (Larsson et al., 2015).It is, however, difficult to define refractive errors in children of varying ages present in our study cohort, but the percentage of children with notable astigmatism stood out, a finding that is rather uncommon in healthy children.Manifest strabismus has been reported in about 3% of healthy children (Larsson et al., 2015;Zhang et al., 2021), which is also considerably less than the 16% noted among our TSC patients.To the best of our knowledge, this is the first study to report on refractive problems and strabismus in paediatric TSC patients specifically; an important matter, especially in children undergoing visual development.CVI may also contribute to poor visual function, having a considerable impact on daily activities, but none of the patients in this cohort had been investigated regarding this diagnosis; a remarkable finding considering the correlation between TSC and CVI reported in the literature (Wan et al., 2019).Diagnosing CVI can be challenging in individuals with severe cognitive or neuropsychiatric problems but cannot explain the total lack of diagnostic work-up for this diagnosis among our TSC patients.Individuals with CVI can benefit greatly from compensatory strategies taught by a specialized vision therapist to facilitate schoolwork and daily life.
Of the 10 patients who underwent ERG due to treatment with vigabatrin, retinal dysfunction was found in three cases, leading to a cessation of the treatment in two children and in the third child they refrained from adjusting the dose with increasing age which could be considered an indirect dose reduction.The level of retinal toxicity is in accordance with the earlier reported 34% risk of visual field loss in children on vigabatrin (Maguire et al., 2010;Riikonen et al., 2015).We have analysed a relatively small number of patients, but the results confirm the importance of monitoring retinal function during treatment with vigabatrin.At follow-up examinations, the 30 Hz flicker response seemed to recover, at least to the level of the initial readings, in the two children in which the vigabatrin treatment was terminated.This was an unexpected finding as the experience in adults is the opposite, where the damage is considered permanent (Kjellström et al., 2008).
A general limitation of the current study is the retrospective design.There is also the possibility of interindividual differences between ophthalmologists' descriptions and awareness of the findings due to the lack of structured study protocols during examinations.Another limitation is the large age range at the final examination, which complicates the comparison of visual acuity as well as refractive errors between different study subjects.A strength with this study was that all individuals with a diagnosis of TSC in the southern health care region of Sweden were included, avoiding selection bias from certain centres.
Most of the manifestations associated with TSC were found in our cohort, apart from CVI.The fact that none of them was examined to detect CVI may, in part, be due to their difficulty in participating in examinations and describing their visual function, but probably also because ophthalmologists are not generally aware of the importance of diagnosing CVI in these cases.It is important that we improve awareness of the risk of CVI among children with TSC.A large proportion of patients in this study had refractive errors and strabismus, and needed spectacles, demonstrating the necessity for extensive ophthalmological examinations of patients with TSC to help their visual development during childhood.

T A B L E 3
Laterality, total number and location of the retinal hamartomas in the 16 tuberous sclerosis complex patients in which hamartomas were identified during any of the ophthalmological examinations.Fundus photograph (a) of an astrocytic hamartoma where the identification can be facilitated by a handheld optical coherence tomography (b, c) (Envisu C2300 VHR SDOIS System).arrows point to the astrocytic hamartoma.

F
Fundus photograph of achromatic patches (white arrows) in a child with tuberous sclerosis complex.T A B L E 4 ffERG recordings for the three patients with reduced responses during vigabatrin treatment.

for a reduced visual acuity Better eye: 18 of 38 patients (%) Worse eye: 25 of 39 patients (%)
Reason or contributing factor to a reduced visual acuity in the patients where monocular visual acuity tests had been possible to perform.
T A B L E 1Note: The eye with the highest level of refractive error measured in the child was selected.Mean age in the 46 children was 12.8 (SD 7.0) years.Abbreviations: D, dioptres; n, numbers.
Patients 1 and 2, examined with the same ffERG Espion system every time, had a reduced amplitude of 30 Hz flicker during treatment with Vigabatrin and a recovery after cessation.In patient 3, the examinations were done with different ffERG systems, making interpretation more difficult but a tendency of reduced responses over time was observed.