Visual Function and Inner Retinal Structure in Relation to Birth Factors in Autosomal Dominant Optic Atrophy

Abstract

Purpose: The extreme variation in expressivity of autosomal dominant optic atrophy (ADOA) is unexplained. It is present from early childhood, why there is reason to search for pre- and perinatal risk factors for poor vision in ADOA. The process of ganglion cell pruning in the fetus is of interest because mitochondria are involved in apoptosis. We hypothesized that suboptimal mitochondrial function makes the developing retina and optic nerve vulnerable to fetal stress in ADOA. We have examined visual function and inner retinal layer structure in relation to birth parameters in ADOA.

Methods: The study included 142 participants with OPA1 ADOA, 62 unaffected first-degree relatives, and 90 unrelated control subjects. Outcome measures included best-corrected visual acuity, microperimetric sensitivity, nerve fiber layer (NFL) volume, and ganglion cell layer (GCL) volume. Descriptive parameters included birth weight, maternal age at birth, birth complications, and gestational age. Analysis was made using mixed modeling.

Results: The analysis showed a significant positive association between microperimetric sensitivity and longer gestational age in ADOA (0.5 dB/week, P = 0.017). Interaction analysis showed a significant different association between microperimetric sensitivity and gestational age between participants with ADOA and the control groups (P = 0.007) and a significant difference in association between NFL volume and birth weight (P = 0.04) and gestational age (P = 0.02) between variant types.

Conclusions: The study suggests that gestational age and birth weight may affect the expressivity of ADOA. The results support that prospectively collected pre- and perinatal data should be included in future studies of the natural history of ADOA.

Autosomal dominant optic atrophy (ADOA) is an inherited disease related to pathogenic variants in the OPA1 gene on chromosome 3, which codes for a mitochondrial dynamin-like GTPase that promotes mitochondrial network stability, bioenergetic output, and control of apoptosis.¹ The ganglion cell and nerve fiber deficit seen as a pale disc and thinning of the inner retinal layers in ADOA (Fig. 1) is highly symmetrical within the two eyes of an individual, but its severity and progression with age are highly variable between patients, even between family members with identical variants. This suggests an influence of additional characteristics or trigger events. 

Figure 1.

 

Fundus photograph (left) and horizontal transfoveal optical coherence tomogram (right) from the left eyes of a healthy individual (above) and an individual with autosomal dominant optic atrophy (below).

During fetal development, the number of primordial ganglion cells is larger than the number present at birth. The reduction is attributed to a natural process of organ development regulated by apoptosis, that is particularly active from the 14th to 30th week of gestation.² We hypothesized that OPA1 insufficiency might increase the vulnerability of retinal ganglion cells to neurotoxic insults during fetal or perinatal life by increasing apoptotic cell death. This study was designed to test this hypothesis by examining the relation between pre- and perinatal exposure to intrauterine growth retardation, perinatal stress, including hypoxia, and the expressivity of ADOA later in life. 

Methods

The study included 3 recruitment groups: 142 participants with genetically verified ADOA of mean age 43 (SD ± 19.7) years (72 men and 70 women) and 2 control groups, namely one with 62 non-carrier first-degree relatives of mean age 38 (SD ± 19.4) years (24 men and 38 women) and one with 90 healthy unrelated controls of mean age 42 (SD ± 18.7) years (40 men and 50 women). The participants were part of a study cohort described elsewhere³ except for two ADOA group participants who underwent genetic testing and joined the cohort recently. Eight participants whose mothers did not consent to the use of birth data were excluded from the present analysis, of which five had ADOA, one was a first-degree relative, and two were unrelated control subjects (Fig. 2). 

Figure 2.

 

Study recruitment flow chart with reference to earlier cohort report.³

Genetic analysis covered 18 pathogenic OPA1 variants known to be found in the families. A founder variant (c.2826_2836delinsGGATGCTCCA) was present in 40% of the participants. Pathogenic variants were categorized as truncating, splice, or other variants.⁴ 

The full clinical examination has previously been described.³ Study examinations included best-corrected visual acuity (BCVA) measured using Early Treatment of Diabetic Retinopathy Study (ETDRS) charts. Macular optical coherence tomography (OCT) was made on Heidelberg Spectralis (HRA + OCT Spectralis OCT2, HRA2 software 6.12.3.0 and HRA + OCT Spectralis OCT1, HRA2 version 6.9.4.0, software version 6.9a; Heidelberg Engineering, Heidelberg, Germany) including a 30 × 25 block of 61 high speed B-scans with averaging of 25 scans. Macular nerve fiber layer (NFL) volume and ganglion cell layer (GCL) volume was retrieved using the manufacturer's automated segmentation algorithm (Heidelberg Eye Explorer, version 1.10.4.0; Heidelberg Engineering, Heidelberg, Germany). One participant, whose scans were deemed not to be reliably segmented, was excluded from analysis. Retinal light sensitivity threshold was measured using the MAIA microperimeter (software version 2.5.1; Centervue, Padova, Italy). The standard protocol with 37 stimuli points covering a circular field ranging from 0 to 10 degrees eccentricity with stimuli at 0, 1, 3, and 5 degrees of eccentricity was centered at the participants’ preferred retinal locus (PRL) of fixation, regardless of whether this was at the center of the anatomic center of the fovea or elsewhere. 

The Danish National Archives (Rigsarkivet) provided digitized handwritten midwife journals for participants born before 1967, digitized birth reports from The Danish Health Authority (Sundhedsstyrelsen) for participants born 1968 to 1988, digitally entered data for participants born 1989 to 1996 from the National Birth Register (Det Medicinske Fødselsregister), and from the National Patient Register (Landspatientregisteret) for participants born in 1997 or later. Additional birth data were provided by Statistics Denmark (Danmarks Statistik) and, for participants born in 1996 or later, from the Danish Medical Birth Register (Det Medicinske Fødselsregister). For participants born between 1994 and 2008, maternal use of prescription medications was pulled from the Register of Pharmaceutical Sales (Lægemiddelstatistikregisteret). The range and quality of data varied with the year of birth of the participant. 

Thus, handwritten midwife journals from 26 participants were interpreted and transcribed with the assistance of a retired senior midwife with personal experience from the era of handwriting. 

A unique personal identification number (CPR number) assigned to all citizens of Denmark was used to link data between registers and between participants and their mothers. 

Study data included birth weight, maternal age at birth, birth complications, maternal medication during birth, APGAR score 5 minutes after delivery, birth presentation, maternal smoking status during pregnancy, and gestational age. Birth weight is expressed in rounded intervals of 100 grams. Maternal age was the mother's age at the time of birth in years. Birth complications were entered as categorical (present or not), including data on prolonged pregnancy, bleeding before or during birth, artificial initiation of labor, cardiac arrhythmia or asphyxia in the child, cesarean section, abnormal birth presentation, labor dystocia (slow or absent progression of labor), pre-eclampsia, high or low birth weight, prematurity, and jaundice. Data on maternal medication during delivery included data on whether the mother received any of the following medication types during birth: chloroform, nitrous oxide with oxygen, trichloroethylene, lidocaine, or any medication in the following categories: general anesthesia, local anesthesia, labor-inducing medications, uterine contracting medications, and spasmolytics. The APGAR score was the one recorded 5 minutes after delivery. Birth presentation was classified as head or breech. Maternal smoking status was categorical (did or did not smoke during pregnancy). Gestational age was counted in weeks. 

All data were collected with consent from both the participant and his or her mother. All mothers who were alive, capable of consenting, and reachable were invited to provide written consent to the retrieval and use of their data. Nine mothers were unreachable as the participant was adopted, or the mother was incapable of cooperating, and those participants were excluded from analysis, leaving a total of 296 participants. For deceased mothers, the Danish Patient Safety Authority authorized data access (3-3013-2678/1). The study was approved by the local Medical Ethics Committee (H-17017461) and the Danish Data Protection Agency (RH-2017-319, I-suite 05947/P-2020-704). The study adhered to the tenets of the Declaration of Helsinki. 

Statistical analysis in R Studio version 4.2.1 included mixed modeling using the nlme package (Pinheiro J, Bates D, R Core Team 2022, version 3.1–159) and the broom.mixed package (Bolker B, Robinson D 2022, version 0.2.9.4) with family and unique study ID as a random effect. Data from both eyes were included, and the models were corrected for age and sex in relevant analyses. The number of observations in some subgroups were low due to lack of data. Only subgroups with more than three observations per group were included in tables and statistical analysis in accordance with General Data Protection Regulation (GDPR) rules set by Statistics Denmark. Variables with fewer than three observations per group are only referred to in the text below, not in the tables. This applies to APGAR scores, use of medication during delivery, birth presentation, and maternal smoking status. Participants were categorized in the ADOA group if they had a known pathogenic variant, regardless of whether they did or did not have the ADOA phenotype at the time of examination. The level of statistical significance was set to 0.05. 

Results

Participants with ADOA, non-carrier first-degree relatives and unrelated controls were of comparable age and gender distribution, whereas mean BCVA was 30 and 31 ETDRS letters lower and of far wider range in participants with ADOA than in their first-degree relatives and the unrelated control group, respectively (P < 0.0001; Table 1). In ADOA, mean NFL volume was 59% and 62% lower and mean GCL volume 56% and 58% lower than in the respective control groups (P < 0.0001). Average microperimetric sensitivity was significantly lower in participants with ADOA than in both control groups (P < 0.0001). Axial length in ADOA was longer by 0.4 mm (P = 0.4) and 0.7 mm (P = 0.006), compared to related and unrelated controls, respectively, whereas intraocular pressure was comparable in the 3 recruitment groups (P > 0.05). Maternal age was 1.7 years lower in ADOA than in the unrelated control subjects (P = 0.02). Birth weight, birth complications, and gestational age were comparable among the three recruitment groups. The APGAR score (range 0–10, from stillborn to perfectly healthy) covered a range from 6 to 10, with 125 participants with ADOA or 94% of evaluable cases obtaining the maximum score of 10 at 5 minutes after delivery. Birth presentation was the head in 121 (94% of evaluable cases) and breech in 4 participants. A history of the mother not having smoked tobacco during pregnancy was obtained in 66 participants (80% of evaluable cases). Exposure to delivery-related medication occurred in 141 participants (81% of evaluable cases). No significant difference was found between recruitment groups for any of these non-tabulated factors (P > 0.05). 

Table 1.

 

Characteristics of Participants With ADOA, Non-Carrier First-Degree Relatives, and Unrelated Control Subjects

 

Analysis of birth parameter effects by OPA1 variant type (truncating, splice, or other; Table 2) found comparable distributions of maternal age, but imbalances in birth weight (splice versus other, P = 0.04), birth complication frequency (splice versus truncating, P = 0.02), and gestational age (splice versus other, P = 0.04). Thus, splice variants had the highest frequency of birth complications and the lowest birth weight. No overall comparison among the three variant groups was significant (see Table 2). 

Table 2.

 

Birth Parameters in Different Variant Types of OPA1 ADOA

 

An exploratory analysis of association between outcome parameters and birth parameters showed a 0.5 decibel (dB) higher microperimetric sensitivity per 1 week longer gestational age (P = 0.0017; Table 3). No other significant associations were identified. 

Table 3.

 

Relation of Visual Acuity With Nerve Fiber Layer Volume, Ganglion Cell Layer Volume, and Birth Parameters in Participants With OPA1 Variants

 

An interaction analysis of associations between outcome parameters and birth parameters by recruitment group is shown in Table 4. A significant association was found between microperimetric sensitivity and gestational age for participants with ADOA (0.56 dB/week, P = 0.0001) in accordance with similar results in Table 3. A significant difference in the association between microperimetric sensitivity and gestational age was found among all three groups (P = 0.007) and among participants with ADOA and the unrelated controls (not tabulated, P = 0.03). All other analyses yielded no significant difference in association (P > 0.05). 

Table 4.

 

Comparison of Associations Between Outcome Parameters and Birth Parameters in Participants With ADOA, Non-Carrier First-Degree Relatives and Unrelated Control Subjects

 

A similar interaction analysis of differences in association between outcome parameters and birth parameters by variant group is shown in Table 5. A significant association was found between microperimetric sensitivity and gestational age for splice mutations (0.76 dB/week, P = 0.008). Overall comparison analysis yielded a significant difference in association between NFL volume and birth weight (P = 0.04) and gestational age (P = 0.02) among the three variant groups. A detailed analysis comparing the variant groups (not tabulated) showed a difference in association between NFL volume and birth weight (not tabulated, P = 0.02) and gestational age (not tabulated, P = 0.02), respectively, between splice and other variants. It also yielded a difference in the association between NFL volume and gestational age between truncating and other variants (not tabulated, P = 0.03). All other differences in associations tested were nominally insignificant (P > 0.05). 

Table 5.

 

Comparison of Association Between Outcome Parameters and Birth Parameters in Different Variant Types of OPA1 ADOA

 

Discussion

Autosomal dominant optic atrophy was associated, as expected, with poorer microperimetric sensitivity than in healthy subjects, but among participants with ADOA, microperimetric sensitivity increased with increasing gestational age, suggesting that the fetal environment benefits the development of vision in ADOA. Poorer microperimetric sensitivity has also been observed in relation to preterm birth,^5,6 and is associated with poorer academic achievement.⁷ Intrauterine growth retardation has also been associated with impaired visual function.⁸ In this context, it is not unreasonable to assume that ADOA may confer an increased vulnerability to perinatal stress and a benefit from delivery at a later date. Benefit of longer gestational age on visual function, in the form of higher microperimetric sensitivity, has also been found in studies of retinopathy of prematurity, where it appeared to be an effect of prematurity as such and not an effect of retinopathy of prematurity or treatment thereof.^9,10 Nevertheless, extremely preterm children have thinner inner retinal layers than term children, which is also the layer where the neuronal defect is located in ADOA.¹¹ In agreement with previous findings, participants with ADOA had poorer visual acuity, microperimetric sensitivity, lower NFL and GCL volume, and longer axial length than healthy subjects.^12–15 

Surprisingly, no post-natal risk factors for neuronal or visual deficit has been found in ADOA, in striking contrast to the vulnerability of people with Leber hereditary optic neuropathy to the consumption of tobacco and alcohol, high IOP, and ethambutol treatment.^16–19 The current study investigated the association between circumstances during pregnancy and birth and the expressivity of autosomal dominant optic atrophy and did not address any particular risk factors. The results suggest that some effects may exist, but they are not strong enough to explain the extremes of outcome variation among family members with identical pathogenic variants. 

The study is large for its field and included only genetically verified cases. This exploratory study without prespecified primary end points was the first to examine visual and structural outcomes in ADOA in relation to birth parameters. It is limited by the retrospective retrieval of birth data. Whereas birth weight and other specific characteristics have been consistently registered, this was not the case for the use of medications, the recording of delivery details, etc. Information about APGAR score, birth presentation, maternal smoking status, and use of medication during delivery was omitted from the analysis due to a low number of observations and low frequencies of certain types of genetic variants. 

The present study does not refute the notion that a disturbance of ganglion cell pruning, which removes nearly half of the primordial ganglion cells and axons in the fetus may affect the expressivity of ADOA.^20,21 The study shows, however, that further studies of the natural history of ADOA can benefit from comprehensive prospective data collection beginning before birth in affected pedigrees. 

Acknowledgments

The authors would like to thank midwives Stinne Høgh from the Neurobiology Research Unit at Rigshospitalet and Lillian Bondo from the Nordic Federation of Midwives, who contributed with their expertise on interpretation of handwriting and assessment of birth register data. Support regarding interpretation from colleagues Mette Vorborg and Helle Josefine Fuchs and her spouse Lars Franch Andersen has been greatly appreciated. We would also like to thank Lene Theil Skovgaard and Brice Ozenne from the Biostatistical Department of Copenhagen University for statistical support. 

Funded by the Synoptik Foundation (Synoptik-Fonden), Fight for Sight Denmark (Øjenforeningen Værn om Synet), The Danish Eye Research Foundation (Øjenfonden) and additionally supported by reimbursements from the regional COVID-19 funds. 

Disclosure: C. Eckmann-Hansen, None; T. Bek, None; B. Sander, None; M. Larsen, Stoke Therapeutics (C)