The Natural History of Leber’s Hereditary Optic Neuropathy in an Irish Population and Assessment for Prognostic Biomarkers

ABSTRACT

In this study we have assessed the clinical and genetic characteristics of an Irish Leber’s hereditary optic neuropathy (LHON) cohort and assessed for useful biomarkers of visual prognosis. We carried out a retrospective review of clinical data of patients with genetically confirmed LHON presenting to an Irish tertiary referral ophthalmic hospital. LHON diagnosis was made on classic clinical signs with genetic confirmation. Alternate diagnoses were excluded with serological investigations and neuro-imaging. Serial logarithm of the minimum angle of resolution (logMAR) visual acuity (VA) was stratified into ‘on-chart’ for logMAR 1.0 or better and ‘off-chart’ if worse than logMAR 1.0. Serial optical coherence tomography scans of the retinal nerve fibre layer (RNFL) and ganglion cell complex (GCC) monitored structure. Idebenone-treated and untreated patients were contrasted. Statistical analyses were performed to assess correlations of presenting characteristics with final VA. Forty-four patients from 34 pedigrees were recruited, of which 87% were male and 75% harboured the 11778 mutation. Legal blindness status was reached in 56.8% of patients by final review (mean 74 months). Preservation of initial nasal RNFL was the best predictor of on-chart final VA. Females had worse final VA than males and patients presenting at < 20 years of age had superior final VA. Idebenone therapy (50% of cohort) yielded no statistically significant benefit to final VA, although study design precludes definitive comment on efficacy. The reported cases represent the calculated majority of LHON pedigrees in Ireland. Visual outcomes were universally poor; however, VA may not be the most appropriate outcome measure and certain patient-reported outcome measures may be of more use when assessing future LHON interventions.

Introduction

Leber’s hereditary optic neuropathy (LHON, OMIM#535000) is an inherited optic neuropathy (typically mitochondrial, rarely autosomal recessive) with a prevalence of 1:30,000–50,000.^1–4 Males are affected in 80–90% of cases (3.4:1 with females, overall male prevalence 1:14,000).^1,2 This is unexplained by mitochondrial inheritance, and likely modified by hormonal factors, mitochondrial deoxyribonucleic acid (mtDNA) haplotype or nuclear DNA modifiers.^1,5–8 Lifetime LHON risk in mtDNA mutation carriers is 50% and 10% for males and females, respectively.⁹

Onset is typically in the second to third decade (95% < 50 years) with sequential involvement over weeks to months (97% bilateral involvement within 1 year).⁹ Classic presentation is painless (sub-)acute sequential visual loss (centrocaecal scotomata) with a hyperaemic optic nerve heads (ONH) and telangiectatic vessels; however, the ONH may often appear normal.¹⁰ Visual prognosis is poor with legal blindness in most cases.¹¹ Syndromic features include: a neurological multiple sclerosis-like syndrome (i.e., Harding’s disease, particularly females); dystonia; and cardiac conduction abnormalities, which may present with ocular involvement or in isolation.^12–15

Three mtDNA mutations (ND4/11778, ND1/3460, ND6/14484) account for >90% of cases, with 11778 being most prevalent (>70%); however, 16 other pathogenic mtDNA mutations and certain nuclear gene loci have been reported.^4,16–20 All known pathogenic mtDNA mutations affect complex I of the mitochondrial respiratory chain reducing adenosine triphosphate (ATP) production and accelerating reactive oxygen species (ROS) generation.⁷ This ATP deficiency disables active axonal transport within the retinal ganglion cells (RGC) and the ROS surplus from anaerobic metabolism contributes to RGC death. Disrupted mitochondrial calcium homoeostasis may predispose to ‘glutamate excitotoxicity’ and RGC apoptosis.^9,21 The retinal nerve fibre layer (RNFL) of the papillomacular bundle (PMB) has a long unmyelinated segment exquisitely sensitive to these deleterious metabolic conditions.²² PMB-RNFL cell dysfunction causes the classic centrocaecal scotoma of LHON and these sustained hostile metabolic conditions may lead to PMB RGC death. A subset of melanopsin-containing RGCs are resistant to mitochondrial dysfunction, preserving normal pupillary light reflexes.^23,24 Unaffected carriers tend to have higher mitochondrial biomass in RGCs than LHON probands, which may confer a protective advantage, but this is not easily assessed in vivo.²⁵

This paper reports real-world natural history data from an Irish LHON cohort. A secondary aim was to assess for any biomarkers at presentation that might predict final functional outcome.

Methods

The study was based on a retrospective audit and complied with local general data protection regulation guidelines (articles 6 & 9). All patients attended the Neuro-Ophthalmology department of the Royal Victoria Eye & Ear Hospital, Dublin, Ireland. mtDNA-positive cases were identified via laboratory and clinic registers. A clinical diagnosis of LHON was confirmed by characteristic symptoms and signs. Other optic neuropathies (e.g., multiple sclerosis, neuromyelitis optica, anti-myelin oligodendrocyte glycoprotein antibody-positive optic neuropathy) were excluded by extensive investigations including neuro-imaging, electrodiagnostics, serum tests and neurological workup, including lumbar puncture with opening pressure and cerebrospinal fluid (CSF) analysis as appropriate.

Patients were clinically followed with serial logarithm of the minimum angle of resolution (logMAR) visual acuity (VA), slit lamp biomicroscopy, kinetic visual fields (Octopus Visual Field, Haag-Streit, Switzerland), spectral domain optical coherence tomography (OCT) (Carl Zeiss HD-OCT 5000, Dublin, CA, United States of America [USA]) and visual evoked potentials (VEP). RNFL and ganglion cell complex (GCC) thickness were measured from OCT, as previously described.²⁶ VA was divided into ‘on-chart’ and ‘off-chart’ categories for those able to read 1.0 or better and worse than 1.0 logMAR, respectively. For ‘off-chart’ VA, logMAR 2.1, 2.4 and 2.7 were recorded for counting fingers, hand movement perception and perception of light, respectively.²⁷ All VA metrics stated in this paper are in the logMAR format.

Genetic testing was performed externally by accredited laboratories in Ireland and the United Kingdom (UK) (Beaumont Hospital, Dublin, Ireland; Oxford Molecular Genetics Laboratory, Oxford, UK; Mitochondrial Biology Unit, Cambridge, UK). Direct sequencing of mtDNA from venous blood was performed for the three main mutations with further screening when these mutations were absent.^17,28

Reimbursement of idebenone (Raxone [R]) for the treatment of LHON in Ireland was first approved in June 2019. Any patients treated with this drug prior to that time had to purchase it privately from the USA. Eligibility criteria for reimbursement included ‘onset of visual loss in the most recently affected eye of ≤ 5 years’.²⁹ Idebenone is funded through the high-tech drug scheme and only available on prescription from one specialist prescriber in Ireland (LC). Treatment is not recommended beyond 2 years if there has been no response within this time period.

Anonymised data were initially stored, processed, and tabulated in a Microsoft Excel workbook. Statistical analysis was undertaken using Minitab 17, Minitab 20 (Minitab LLC, Pennsylvania, USA), Statistical Product and Service Solutions (SPSS) 25 (IBM Corporation, New York, USA), and Microsoft Excel. Hypothesis tests used included t-tests, chi-square tests of independence, and analysis of variance (ANOVA). Scatterplots and R² were used to determine approximate correlations in linear regression analyses. The Benjamini Hochberg procedure was used to correct the false discover rate (FDR) for overall multiple comparisons, and the Tukey method was used to correct the FDR within ANOVAs. All p-values reported are post FDR correction and a value of ≤ .05 was presumed to be statistically significant. All logs and data files have been saved for future reference.

Results

NB The tables hereunder do not always represent the entire cohort as a subset of data was available for analysis under each interrogated parameter, due to the retrospective nature of the study.

Demographics & general findings (Table 1)

Table 1.

Age of onset and errata

		Age of onset (years ± SD)	Age of onset; difference between groups?	$\chi$ 2 gene type 11778 versus 14484/3460/15257
Gender	Female Male	38.50 ± 13.63 (n = 6) 26.66 ± 14.77 (n = 32)	t-test t = 2.58 p = .082	$\chi$ 2 = 1.126 p = .456
Family history	No Yes	29.95 ± 16.37 (n = 19) 27.11 ± 13.88 (n = 19)	t-test t = 0.82 p = .809	$\chi$ 2 = 7.472 p = .045
Gene type	11778 14484/3460/15257	25.89 ± 12.96 (n = 28) 35.90 ± 18.46 (n = 10)	t-test t = 2.64 p = .080	N/A
Disc appearance	Telangiectasia Swollen Optic atrophy Normal	24.78 ± 9.99 (e = 18) 23.57 ± 11.38 (e = 14) 33.09 ± 17.78 (e = 22) 30.71 ± 18.13 (e = 14)	1-way ANOVA F = 1.67 p = .455	$\chi$ 2 = 1.906 p = .732
Electrical response to pVEP?	No response Response	30.94 ± 15.96 (e = 16) 30.85 ± 19.05 (e = 20)	t-test t = 0.01 p = .998	$\chi$ 2 = 0.900 p = .486
Plasmy	Heteroplasmy Homoplasmy	31.50 ± 17.90 (n = 2) 28.20 ± 12.38 (n = 15)	t-test t = 0.48 p = .848	*
MRI	Normal Abnormal	26.50 ± 13.18 (n = 22) 35.50 ± 22.42 (n = 6)	t-test t = −1.78 p = .650	$\chi$ 2 = 2.233 p = .312

all p-values are post-false discover rate correction

e = number of eyes

MRI = magnetic resonance imaging

n = number of patients

pVEP = pattern visual evoked potentials

SD = standard deviation

*= insufficient data

NB Only patients with all required criteria per comparison are represented in each subsection, thus the total number for each category may not always equal 44.

Between 1988 and 2020, 44 patients from 34 pedigrees had genetically confirmed LHON with 84% being male (n = 37). Review duration (for 73% with multiple visits) was for a median 33.5 months (mean 74.4, range 2.0–431.2 months); 84.4% (n = 27/32) of the patients had follow up for ≥ 6 months. Presentation was simultaneous in 27.3%, sequential in 45.5% with inadequate data in 27.3%. Mean time to fellow eye involvement in sequential cases was 5.5 standard deviation (SD) 7.5 (median 3) months. Time to presentation from symptom onset data were available for 22 patients. Mean time to presentation was 3 months, (range 0–13 months), with 32% of patients presenting within 1 month of symptom onset. Mean age of onset was 28.5 SD 15.1 years. Earlier age of onset was non-significantly associated with male gender (26.7 years for males versus 38.5 years for females, p = .082), and the 11778 mutation (p = .080). Presenting age was <20 and ≥20 years in 39%, and 61%, respectively. Family history was present in 43.2% (n = 19/44) and was associated with having the 11778 mutation versus other mutations (χ2 = 7.472, p = .045). Affected males, females and mothers were found in 84.2% (n = 16/19), 31.6% (n = 6/19) and 10.5% (n = 2/19) of cases with a positive family history, respectively.

The most prevalent mtDNA mutations were 11778 (75%), 14484 (15.9%), 3460 (6.8%) and 15257 (2.3%), with 11778 being causative in 86% of females (n = 6/7) and 73% of males (n = 27/37). Owing to testing methodology, only 41% (18/44) of patients had an assessment of plasmy, of which 11% (2/18) were heteroplasmic with a mean 44% wild type mtDNA. ONHs of affected eyes at presentation showed 28.6% telangiectasia, 22.9% swelling, 28.6% optic atrophy and 20% normal features.

Systemic features were present in 11.36% (5/44), manifesting as Harding’s disease in three cases (2x 11778 and 1 × 3460 mutations), cardiac features in one case (partial left bundle branch block with the 3460 mutation) and one case of generalised dystonia (with the 14484 mutation). Magnetic resonance imaging (MRI) results were available for 64% (28/44) with abnormalities detected in six cases (four with optic nerve enhancement, one with optic atrophy and one with both). Genetic aetiology for positive MRIs was 11778 (n = 3), 14484 (n = 2) and 3460 (n = 1). Abnormal findings on MRI were non-significantly associated with a later age of onset (35.5 versus 26.5 years, p = .650).

Visual acuity (Table 2)

Table 2.

Visual acuity data

		Presenting VA (logMAR ± SD)	Presenting VA; difference between groups?	Final VA (logMAR ± SD)	Final VA; difference between groups?	$\chi$ 2 versus response to Idebenone Y/N?	$\chi$ 2 versus legally blind status	$\chi$ 2 vs ≥ 15 letter improvement
Gender	Female Male	0.904 ± 0.717 (n = 6) 1.030 ± 0.860 (n = 28)	t-test t = −0.47 p = .848	2.060 ± 0.465 (n = 5) 1.55 ± 0.774 (n = 31)	t-test t = 2.02 p = .198	*	$\chi$ 2 = 0.610 p = .567	$\chi$ 2 = 2.094 p = .317
Family history	No Yes	1.074 ± 0.773 (n = 17) 0.943 ± 0.896 (n = 17)	t-test t = 0.64 p = .842	1.405 ± 0.739 (n = 20) 1.891 ± 0.701 (n = 16)	t-test t = −2.83 p = .073	$\chi$ 2 = 0.953 p = .486	$\chi$ 2 = 3.783 p = .223	$\chi$ 2 = 0.005 p = .968
Age of onset	< 20 ≥ 20	0.912 ± 0.790 (n = 13) 1.068 ± 0.863 (n = 21)	t-test t = −0.75 p = .822	1.326 ± 0.858 (n = 17) 1.884 ± 0.543 (n = 19)	t-test t = −3.33 p = .016	$\chi$ 2 = 2.723 p = .427	$\chi$ 2 = 3.849 p = .223	$\chi$ 2 = 3.389 p = .230
Gene type	11778 14484/3460 /15257	1.015 ± 0.831 (n = 25) 0.989 ± 0.862 (n = 9)	t-test t = −0.11 p = .947	1.654 ± 0.775 (n = 26) 1.535 ± 0.722 (n = 10)	t-test t = −0.59 p = .847	$\chi$ 2 = 3.020 p = .246	$\chi$ 2 = 0.004 p = .968	$\chi$ 2 = 0.002 p = .968
Disc appearance	Telangiectasia Swollen Optic atrophy Normal	0.978 ± 0.875 (e = 18) 0.700 ± 0.628 (e = 14) 1.372 ± 0.816 (e = 16) 0.864 ± 0.850 (e = 14)	1-WAY ANOVA F = 1.92 p = .136	2.072 ± 0.325 (e = 18) 1.293 ± 0.878 (e = 14) 1.745 ± 0.652 (e = 22) 1.340 ± 0.836 (e = 10)	ANOVA F = 4.52 p = .012	$\chi$ 2 = 6.050 p = .297	$\chi$ 2 = 13.581 p = .045	$\chi$ 2 = 12.683 p = .045
Plasmy	Hetero Homo	1.500 ± 0.693 (n = 2) 1.061 ± 0.885 (n = 14)	t-test t = 0.95 p = .793	0.875 ± 0.675 (n = 2) 1.764 ± 0.631 (n = 14)	t-test t = −2.62 p = .092	$\chi$ 2 = 9.079 p = .045	$\chi$ 2 = 1.524 p = .402	*
Idebenone	No Yes	1.137 ± 0.766 (n = 13) 0.929 ± 0.872 (n = 21)	t-test t = 1.00 p = .534	1.584 ± 0.860 (n = 16) 1.650 ± 0.674 (n = 20)	t-test t = −0.36 p = .793	N/A	$\chi$ 2 = 2.363 p = .310	$\chi$2 = 1.422 p = .401

all p-values are post-false discover rate correction

e = number of eyes

logMAR = logarithm of the minimum angle of resolution

n = number of patients

SD = standard deviation

VA = visual acuity

*= insufficient data

NB Only patients with all required criteria per comparison are represented in each subsection, thus the total number for each category may not always equal 44.

Mean logMAR VA (n = 34 patients, 68 eyes) was 1.01 SD 0.83 at presentation and 1.62 SD 0.76 at final review, with a change of +0.68 SD 0.99 over the study period. Despite better presenting VA (0.90 vs 1.03, p = .848), females ended with worse final VA than males (2.06 versus 1.55, p = .198). A positive family history (1.89 versus 1.41, p = .073), telangiectatic or atrophic discs on presentation (2.07 (telangiectasia)/1.75 (atrophic) vs 1.29 (swollen)/1.34 (normal), p = .012), and homoplasmy (1.76 versus 0.88, p = .092) were associated with worse final VA. Normal or swollen optic discs were more associated with ≥ 15 letter spontaneous improvement than telangiectatic or atrophic discs (χ2 = 12.683, p = .045). There were no statistically significant associations between mtDNA genotype and presenting (p = .947) or final (p = .847) VA; however, final VA was superior in heteroplasmic patients (0.88 versus 1.76, p = .092), though limited quantification of heteroplasmy prevented assessment of direct correlation with the magnitude of VA difference. Age of onset < 20 years was significantly associated with better final VA than ≥ 20 years (1.33 versus 1.88, p = .016), the equivalent of 27.9 letters or a 5.58-line improvement. VA was ‘on-chart’ (i.e., numerically ≤ 1.0 logMAR) in 58.8% (n = 40/68)   at presentation (mean VA 0.33 SD 0.33) but only 27.9% (n = 19/68)   at final review (mean VA 0.46 SD 0.27). Final VA was better for eyes presenting with ‘on-chart’ VA (1.45 vs 1.97, p = .068).

The legal requirements for driving a non-commercial vehicle (i.e., logMAR 0.3 or better in ≥ 1 eye) were met by 43.2% (n = 19/44) of patients initially, but only 11.4% (n = 5/44) at last review. The legal blindness criterion (i.e., logMAR 1.0 or worse in the better eye) was met in 15.9% (n = 7/44) of patients at presentation and 56.8% (n = 25/44) at final review. Telangiectatic or atrophic discs at presentation were more associated with legal blindness than normal or swollen discs (χ2 = 13.581, p = .045), and the reverse associations were true with respect to ≥15 letters of improvement (χ2 = 12.683, p = .045).

OCT data (Table 3)

Table 3.

Selected optical coherence tomography data

Subgroups		1st OCT average RNFL thickness (µm ± SD)	1st OCT average RNFL thickness; subgroup difference?	Final OCT average RNFL thickness (µm ± SD)	Final OCT average RNFL thickness; subgroup difference?	1st OCT average GCC thickness (µm ± SD)	1st OCT average GCC thickness; subgroup difference?	1st OCT nasal RNFL thickness (µm ± SD)	1st OCT nasal RNFL thickness; subgroup difference?	1st OCT superior RNFL thickness (µm ± SD)	1st OCT superior RNFL thickness; subgroup difference?
Gender	Female Male	86.44 ± 7.26 (n = 2) 88.15 ± 28.94 (n = 20)	t-test t = −0.23 p = .947	90.4 ± 23.5 (n = 1) 73.9 ± 16.1 (n = 7)	t-test t = 1.29 p = .664	74.42 ± 3.04 (n = 2) 55.97 ± 11.95 (n = 10)	t-test t = 7.30 p < .001	62.50 ± 5.74 (n = 2) 71.8 ± 23.9 (n = 10)	t-test t = −1.54 p = .489	105.50 ± 6.40 (n = 2) 108.2 ± 42.0 (n = 10)	t-test t = −0.27 p = .942
Gene type	11778 14484/3460 /15257	90.36 ± 26.60 (n = 10) 75.4 ± 25.1 (n = 2)	t-test t = −1.04 p = .793	75.52 ± 18.51 (n = 7) 79.63 ± 3.36 (n = 1)	t-test t = 0.73 p = .845	57.43 ± 12.05 (n = 17) 60.25 ± 16.84 (n = 3)	t-test t = 0.50 p = .847	73.25 ± 22.98 (n = 10) 55.50 ± 6.24 (n = 2)	t-test t = −2.95 p = .075	111.90 ± 37.89 (n = 10) 87.0 ± 38.0 (n = 2)	t-test t = −1.20 p = .702
Plasmy	Heteroplasmy Homoplasmy	86.2 ± 33.6 (n = 2) 84.19 ± 16.06 (n = 8)	t-test t = 0.12 p = .947	76.2 ± 21.2 (n = 2) 76.0 ± 16.7 (n = 6)	t-test t = 0.02 p = .992	58.13 ± 12.99 (n = 2) 63.19 ± 13.01 (n = 10)	t-test t = −0.71 p = .842	70.0 ± 17.7 (n = 2) 63.44 ± 8.49 (n = 8)	t-test t = 0.72 p = .842	115.8 ± 55.2 (n = 2) 102.6 ± 22.1 (n = 8)	t-test t = 0.47 p = .856
Age of onset	<20 ≥20	76.9 ± 30.1 (n = 3) 91.5 ± 24.9 (n = 9)	t-test t = −1.18 p = .702	94.4 ± 5.1 (n = 1) 73.3 ± 16.7 (n = 7)	t-test t = 3.59 p = .094	51.09 ± 9.22 (n = 7) 61.3 ± 12.9 (n = 13)	t-test t = −2.52 p = .073	66.0 ± 15.7 (n = 3) 71.7 ± 24.0 (n = 9)	t-test t = −0.54 p = .847	94.2 ± 54.3 (n = 3) 112.3 ± 32.1 (n = 9)	t-test t = −1.00 p = .793
Presenting VA	On chart Off chart	87.56 ± 20.70 (e = 13) 75.22 ± 17.00 (e = 9)	t-test t = 1.47 p = .553	90.1 ± 10.6 (e = 7) 63.8 ± 11.2 (e = 8)	t-test t = 4.67 p = .009	63.14 ± 12.54 (e = 17) 53.51 ± 11.38 (e = 17)	t-test t = 2.34 p = .163	67.1 ± 12.1 (e = 13) 60.67 ± 5.22 (e = 9)	t-test t = 1.70 p = .427	107.8 ± 34.7 (e = 13) 92.6 ± 23.6 (e = 9)	t-test t = 1.14 p = .729
Final VA	On chart Off chart	105.7 ± 16.8 (e = 3) 78.6 ± 19.1 (e = 17)	t-test t = 2.30 p = .150	89.33 ± 9.45 (e = 3) 72.8 ± 17.4 (e = 12)	t-test t = 2.24 p = .303	61.29 ± 9.40 (e = 4) 56.7 ± 12.9 (e = 29)	t-test t = 0.68 p = .842	81.33 ± 7.51 (e = 3) 62.24 ± 8.08 (e = 17)	t-test t = 3.80 p = .016	143.0 ± 35.7 (e = 3) 93.4 ± 26.4 (e = 17)	t-test t = 2.87 p = .080
Idebenone response?	No Yes	79.4 ± 19.2 (n = 6) 85.3 ± 25.3 (n = 4)	t-test t = −0.57 p = .847	81.6 ± 14.8 (n = 4) 75.4 ± 15.5 (n = 4)	t-test t = 0.76 p = .822	56.6 ± 13.5 (n = 10) 60.0 ± 13.4 (n = 4)	t-test t = −0.58 p = .847	63.83 ± 8.82 (n = 6) 66.9 ± 14.3 (n = 4)	t-test t = −0.57 p = .847	93.8 ± 28.0 (n = 6) 111.4 ± 40.6 (n = 3)	t-test t = −1.13 p = .752
MRI	Normal Abnormal	79.25 ± 16.33 (n = 7) 110.19 ± 6.84 (n = 2)	t-test t = −5.58 p = .039	75.4 ± 19.5 (n = 4) 90.69 ± 8.34 (n = 2)	t-test t = −1.81 p = .427	58.24 ± 12.71 (n = 13) 70.92 ± 6.35 (n = 2)	t-test t = −3.14 p = .436	61.00 ± 6.37 (n = 7) 81.75 ± 5.74 (n = 2)	t-test t = −5.85 p = .016	93.6 ± 24.1 (n = 7) 144.8 ± 23.9 (n = 2)	t-test t = −3.74 p = .143

all p-values are post-false discover rate correction

e = number of eyes

n = number of patients

GCC = ganglion cell complex

MRI = magnetic resonance imaging

OCT = optical coherence tomography

RNFL = retinal nerve fibre layer

SD = standard deviation

VA = visual acuity

NB Data represent available OCT scans for patients at each appropriate time point, thus categories may have numbers less than the total cohort (e.g., patient recruited in 1988 does not have OCT data).

OCT scans of GCC and RNFL were performed in 44.32% (20/44 patients, 39/88 eyes) with serial scans in 20.45% (9/44 patients, 18/88 eyes). Mean interval from initial to final OCT was 6.5 SD 3.1 months. Mean initial figures were: RNFL thickness 87.86 SD 26.44 µm (n = 24/88), GCC thickness 57.86 SD 12.67 µm (n = 39/88), and neuroretinal rim (NRR) area 1.27 SD 0.42 mm² (n = 23/88). Mean change in RNFL and GCC thickness, and NRR area were -9.91 SD 12.92 µm (n = 14), −11.48 SD 10.79 µm (n = 17), and −0.134 SD 0.18 mm² (n = 13), respectively. Final RNFL, GCC and NRR measurements were 78.57 SD 14.95 µm, 52.17 SD 7.68 µm, and 1.17 SD 0.31 mm²,respectively.

Males had a significantly thinner initial GCC thickness than females (55.97 versus 74.42 µm, p < .001) true for the mean and all sextants. Having a thinner nasal quadrant RNFL layer at the initial OCT was significantly associated with having an ‘off-chart’ final VA (62.24 vs 81.33 µm, p = .016). For RNFL, a thinner superior quadrant at presentation was non-significantly associated with ‘off-chart’ final VA (93.4 vs 143.0 µm, p = .080). Other quadrants and mean RNFL measurements at presentation were not predictive of final VA.

Idebenone data (Table 2)

Idebenone therapy (900 mg/day) was commenced in 50% (n = 22/44) of patients a median 4 months (range 1–36) after initial presentation and was continued for 2 years. Idebenone therapy was not associated with any significant differences in VA change (+0.661 vs +0.700, p = .968) or final VA (1.584 vs 1.650, p = .793), though heteroplasmy was significantly associated with idebenone response (p = .045). Ten eyes of six patients (22.7%, 10/44 eyes) in the idebenone-treated group showed improvement from presenting VA (mean improvement −0.59 SD 0.35). Responding cases were predominantly 11778 aetiology (83.3%, n = 5/6). Spontaneous visual recovery was detected in three eyes of two patients (6.8%, n = 3/44 eyes) in the untreated group (mean improvement −1.03 SD 1.10, 50% 11778 and 50% 14484). Of all eyes that improved, final VA was 0.82 SD 0.49 and 0.73 AD 1.18 for the idebenone treated and untreated groups, respectively.

VEP data

VEP data at diagnosis were available for 18 patients (35/88 eyes), being pattern reversal VEP (pVEP) for 19/35 and flash VEP (fVEP) for 16/35 (i.e., where pVEP was non-recordable). All patients with recordable pVEP/fVEP had P100 delay. Mean pVEP P100 latency was 134.77 ms SD 25.15 with a mean amplitude of 4.464 µV SD 4.229. No electrical response to pVEP was seen in 46% (n = 16/35). For fVEP, mean P100 latency and amplitudes were 115.32 SD 34.01 ms and 3.878 SD 3.762 µV for 3 Hz (n = 15) and 142.9 ms and 15.5 µV for 9 Hz (n = 1) stimulation. An absent electrical response to pVEP was weakly associated with ‘off-chart’ final VA (${\chi }^2$ = 4.425, p = .175). No other significant associations were observed within the VEP data.

Discussion

Our main findings were that female gender (p = .198), mtDNA homoplasmy (regardless of variant, p = .092), presenting with ‘off-chart’ VA (p = .068), telangiectatic or atrophic discs (p = .012), non-recordable pVEP (p = .175), and thinner nasal RNFL measurements (p = .016) were useful as biomarkers predicting poorer (i.e., ‘off-chart’) final VA. Those presenting < 20 years of age and with ‘on-chart’ VA had better visual outcomes and those with normal or swollen discs were more likely to improve by ≥ 15 letters. Despite this, LHON still has a universally dismal visual prognosis.¹¹ mtDNA heteroplasmy was a useful biomarker of idebenone response, but idebenone treatment was not statistically significantly associated with better visual outcomes. However, one idebenone-treated patient recovered legal driving vision and others reported functional improvement in vision and currently this is the only approved therapeutic option for LHON.

The 34 LHON pedigrees in this cohort are the presumed majority of the Irish LHON cohort (population 4.7 million), comparable with the published 36 Finnish pedigrees (population 5.2 million).^1,30 Harding et al. reported a positive family history in 64% of LHON cases.⁷ Our cohort falls short of this at 45.5%. Family history was associated with the 11778 mutation, perhaps indicating greater penetrance than the other genotypes (χ2 = 7.472, p = .045). LHON-awareness within affected pedigrees and early presentation may explain initial preservation of P100 amplitude; however, final VA was as poor as the other genotypes (p = .847).

While younger patients had significantly better final VA (p = .016), visual function remained poor (i.e., logMAR 1.33 and 1.88 for < 20 and ≥ 20 years, respectively). Preservation of RNFL/GCC may be of more relevance than VA as a marker for future treatment efficacy. Although LHON classically affects males, we found that affected females had poorer visual outcomes in this Irish cohort (logMAR 2.06 versus 1.55, p = .198).

MRI in LHON may detect afferent visual pathway abnormalities, including T2 short tau inversion recovery sequence high signal, gadolinium contrast enhancement and optic atrophy.^16,31,32 Although more typical of MS-related retrobulbar optic neuropathy, MRI abnormalities do not exclude LHON and MRI enhancement may be transient. The 24% (n =6/25) with an abnormal MRI in our cohort had mainly 11778 mutations (n = 5/6) and a trend towards better final VA. This group was associated with significantly better initial preservation of OCT structures (RNFL and GCC), which may reflect a greater disease burden on the central nervous system than the RNFL, at least initially. This again demonstrates that early retention of inner retinal structures may indicate a therapeutic window that rapidly closes.

Phenotype-genotype correlations have been described for the primary three mtDNA mutations and rare variants; however, there is considerable variability in penetrance and severity depending on the degree of heteroplasmy, other mtDNA factors (e.g., modifier mutations (m.14502 T > C modulation of m.11778 G > A) and haplotype J) and nuclear modifiers.^16,33,34 In our cohort, there was no correlation between mutation type and final VA, though preserved nasal RNFL on OCT was associated with ‘on chart’ final VA (p = .016) and mtDNA heteroplasmy was predictive of better VA outcomes (0.88 versus 1.76, p = .092). We aim to comprehensively resequence the mitochondrial genome and relevant nuclear genes of this cohort for a future study.

Spontaneous partial visual recovery has been reported to be more likely in heteroplasmic individuals, especially those with the 14484 (58% – 65%) and m.4171 C > A (52%) mutations than the 11778 (4–26%) mutation, varying with race, though this was not reflected in our cohort.^9,10,19 Recovery most often occurs in the first 1 to 2 years from presentation but is reported up to 6.5 years.^10,19 This improvement may be difficult to quantify with traditional VA charts, rather manifesting as subjective changes (e.g., improved transparency and ‘fenestrations’ in the centrocaecal scotomata thus highlighting the relevance of patient reported outcome measures (PROMs) as alternate therapeutic endpoints for clinical trials in low vision conditions (e.g. LHON, advanced retinal dystrophy).^9,35 The mechanism of recovery is debated, possibly representing restoration of axonal transport or development of functional reserves of mitochondrial biomass. Recovery is most likely if presenting < 21 years (consistent with our experience), less severe vision loss at the nadir, larger optic disc size and having the 14484 mutation.¹⁰ Moon et al. showed preserved NRR was a predictor of visual recovery, likely correlating with preservation of RNFL axons.¹⁰ Preserved NRR area was not significantly associated with better VA outcomes in our cohort when comparing ‘on-chart’ versus ‘off-chart’ final VA (1.61 versus 1.14 mm², p = .143).

ONH telangiectasia, oedema and hyperaemia are considered negative prognostic signs possibly representing axoplasmic stasis in the prelaminar optic nerve, reflecting severity of mitochondrial dysfunction.¹⁰ Perhaps, this severe optic nerve swelling leads to a secondary anterior ischaemic optic neuropathy removing any chance of significant visual/functional recovery with the resolution of acute oedema. This is an area for further investigation in LHON pedigrees. Regardless of the pathophysiology, our cohort partly concurred with this finding, showing significantly worse final VA in telangiectatic/atrophic discs (p = .012). However, we did find that normal/swollen discs had a greater likelihood of spontaneous recovery of ≥ 15 letters (p = .045). Early, ideally pre-symptomatic, therapy may be the key to preventing visual loss. Atrophic discs at baseline likely represent a delay in presentation, with acute inflammatory signs already resolved, leaving a structurally impaired RNFL behind. This is supported by atrophic discs having the least GCC remaining to be lost between presentation and final review.

Genetic counselling in LHON is difficult due to the unexplained male predilection and unpredictable timing of onset. “Unaffected” mutation carriers may have some overlap with the pre-clinical group, as OCT changes in RNFL may be detectable prior to symptom onset though the risk of manifesting disease is low in the > 50 years population.³⁶ Avoidance of cigarette smoking, excessive alcohol consumption and mitochondria-toxic medications (e.g., ethambutol) is advisable in known carriers.^9,37 mtDNA heteroplasmy > 60% may be protective against LHON, though heteroplasmy may differ by anatomical site (e.g., optic nerve vs leukocytes).³⁸ Prevention of vertical transmission is now possibly via certain in vitro fertilisation techniques using donor rather than maternal mtDNA.³⁹

Objective monitoring

OCT is a useful marker of inner retinal structural change in LHON and other diseases.⁴⁰ Progressive thinning of the GCC, RNFL and inner plexiform layer and thickening of the inner nuclear, outer plexiform and outer nuclear layers has been associated with symptom onset and correlates with decline in VA, particularly GCC thinning in the central macula.^26,41 GCC thinning may indicate a predisposition to vision loss in asymptomatic LHON mutation carriers; however, unless mtDNA status is known in advance, OCT is limited to monitoring progression of atrophy or treatment response after symptom onset.³⁶ Barboni et al. reported that OCT may be useful to predict likelihood of recovery (i.e., a thicker more preserved RNFL at presentation was a positive prognostic factor).⁴² We found that nasal quadrant (and, to a lesser degree, superior) RNFL thinning at the initial OCT predicted ‘off-chart’ final VA (p = .016). Our data showed that 11778 patients lost greater mean GCC thickness across the review duration than the other genotypes, which is likely a result of earlier presentation due to family history awareness rather than a distinct, more gradual pathological mechanism. The correlation of thinner mean RNFL at final OCT and ‘off-chart’ VA at presentation (p = .010) may either reflect delay in presentation or more aggressive rapid disease progression. The available data do not differentiate the underlying route to this anatomical/functional outcome.

Treatment

Previous treatment approaches included micronutrient replacement (e.g. vitamin B₁, vitamin B₁₂, folic acid), hyperbaric oxygen, steroids and calcineurin inhibitors (e.g., cyclosporin), none of which have proven efficacy.^43–45 In vitro studies suggest a protective effect of synthetic oestrogens in keeping with reduced female prevalence.⁴⁶ Current treatments include quinones (e.g., idebenone), which promote respiratory chain function in existing mitochondria while other agents (e.g., metformin, glitazones) increase the number of available RNFL mitochondria (i.e., biomass).^29,47–51 The RHODOS trials showed idebenone (900 mg/day) gave benefit for genetically confirmed LHON for up to 30 months in patients < 5 years from onset (particularly with the 11778 or 3460 mutations), thus early genetic confirmation is essential to access this treatment.^29,48,52,53 Half of our cohort were prescribed idebenone, with no significant difference in final vision between the treated and untreated groups. In the idebenone group, 10 eyes did recover, but to a lesser mean value (29.5 letters, 5.9 lines) than the non-treated spontaneously improving eyes (51.5 letters, 10.3 lines, n = 3 eyes). Idebenone is the only available approved treatment for LHON at this time, but efficacy was not demonstrable in this cohort. However, the goal of this retrospective cohort study was to describe the real-world experience of LHON patients and was not designed to test the direct impact of idebenone on disease progression. This report describes the VA, OCT and genetic parameters associated with an Irish LHON cohort.

Novel mitochondrial and nuclear allotopic gene therapy techniques are under investigation.⁵⁴ Human intravitreal gene therapy trials have shown 15-letter gains in 66%, with effects from monocular injection seen bilaterally.^55,56 A 15-letter improvement from a baseline of logMAR 1.0 still leaves a person significantly visually impaired and more relevant PROMs are required (e.g., visual function questionnaire-25, mobility assessment, task-oriented outcomes).³⁵ Stem cells may help maintain existing RGC integrity (e.g., neurotrophic factors) or replace atrophic RGCs (e.g., induced pluripotent stem cell-derived RGCs) though local and cortical integration may hamper efficacy of the latter.^57,58

Heteroplasmy in LHON families (i.e., a subset of genetically normal mitochondria) may decrease penetrance and infer a better prognosis and treatment response.^1,38 Heteroplasmy is more common in de novo (37.5–80%) versus inherited (5%) mtDNA mutations, suggesting a trend towards homoplasmy in subsequent generations.^1,7,59 Blood leukocytes, hair cells, retina and optic nerve may each have different percentages of mutant mtDNA thus blood testing may not accurately represent the degree of mitochondrial dysfunction in RGCs.^60,61 LHON-affected people usually have > 95% mutant mtDNA.⁷ Disease may still manifest in heteroplasmy due to unfavourable mitochondrial haplotype or nuclear modifiers; however, 13.6–19% of unaffected maternal relatives show mtDNA heteroplasmy and 1:300 UK citizens harbour an LHON mutation (1:1000 homoplasmic) without necessarily manifesting disease, thus other modifiers are likely.^7,59,62,63 In our cohort, heteroplasmic patients had less severe visual loss over the study period. Kim et al. reported patients with the m.14459 G >A mutation manifesting LHON in heteroplasmy and neurological manifestations in homoplasmy thus degree of heteroplasmy may determine disease phenotype.¹⁴

Strengths/limitations

Limitations of this study were primarily due to its retrospective nature and lack of an electronic medical record over the 32-year study period. This resulted in several instances of missing data, though all available data were used to assess for individual predictive factors of presenting and final visual function. There may be sampling bias as only cases attending one tertiary referral ophthalmic centre were included, though, as outlined above, this should be representative. The genetic testing methodologies used may have had limited detection rates of known and novel mtDNA mutations and assessment of the degree of heteroplasmy; however, this will be addressed in a planned resequencing study of LHON patients. The main limitation was the small sample size for sub-categorical comparisons. A strength of this study is the robust statistical methodology employed, though significance was often not reached due to low patient numbers due to both the fact that LHON is a rare disease and that relevant data were not available in all cases.

Conclusions

This report details the Irish picture of LHON for the first time. Despite current treatments, VA outcomes in LHON were poor with 56.8% of patients in this cohort being legally blind and only 11.4% achieving legal driving standards at final review.⁶⁴ This is an indicator of the devastating impact of LHON on meeting social and economic needs of those affected. We detected limited biomarkers of visual outcome (i.e., nasal RNFL, normal/swollen discs); however, LHON prognostic factors are incompletely understood, with available treatments at an unsatisfactory level and further studies required. Full clinical and genetic characterisation of the LHON population is critical for accurate diagnosis/prognosis and to access emerging gene-specific therapies. Perhaps, as with other gene therapies for inherited retinal disease, VA is not the best marker of treatment outcome/success and task-oriented PROMs may be more appropriate study endpoints.^35,65 Nasal RNFL thickness on OCT had some predictive value on final VA and may help select candidates for future therapeutics.

Funding Statement

No specific funding was granted for this project.

Declaration of interest statement

No potential conflict of interest was reported by the authors.

Data availability statement

All source data are available in anonymised format from the corresponding author upon reasonable request.