The Genetic Confirmation and Clinical Characterization of LOXL3-Associated MYP28: A Common Type of Recessive Extreme High Myopia

Abstract

Purpose: In previous studies, biallelic LOXL3 variants have been shown to cause autosomal recessive Stickler syndrome in one Saudi Arabian family or autosomal recessive early-onset high myopia (eoHM, MYP28) in two Chinese families. The current study aims to elucidate the clinical and genetic features of LOXL3-associated MYP28 in seven new families and two previously published families.

Methods: LOXL3 variants were detected based on the exome sequencing data of 8389 unrelated probands with various ocular conditions. Biallelic variants were identified through multiple online bioinformatic tools, comparative analysis, and co-segregation analysis. The available clinical data were summarized.

Results: Biallelic LOXL3 variants were exclusively identified in nine of 1226 families with eoHM but in none of the 7163 families without eoHM (P = 2.97 × 10⁻⁸, Fisher's exact test), including seven new and two previously reported families. Seven pathogenic variants were detected, including one nonsense (c.1765C>T/p.Arg589*), three frameshift (c.39dupG/p.Leu14Alafs*21; c.544delC/p.Leu182Cysfs*3, c.594delG/p.Gln199Lysfs*35), and three missense (c.371G>A/p.Cys124Tyr; c.1051G>A/p.Gly351Arg; c.1669G>A/p.Glu557Lys) variants. Clinical data of nine patients from nine unrelated families revealed myopia at the first visit at about 5 years of age, showing slow progression with age. Visual acuity at the last visit ranged from 0.04 to 0.9 (median age at last visit = 5 years, range 3.5–15 years). High myopic fundus changes, observed in all nine patients, were classified as tessellated fundus (C1) in five patients and diffuse choroidal atrophy (C2) in four patients. Electroretinograms showed mildly reduced cone responses and normal rod responses. Except for high myopia, no other specific features were shared by these patients.

Conclusions: Biallelic LOXL3 variants exclusively presenting in nine unrelated patients with eoHM provide firm evidence implicating MYP28, with an estimated prevalence of 7.3 × 10⁻³ in eoHM and of about 7.3 × 10⁻⁵ in the general population for LOXL3-associated eoHM. So far, MYP28 represents a common type of autosomal recessive extreme eoHM, with a frequency comparable to LRPAP1-associated MYP23.

High myopia (HM), a common cause of vision impairment and acquired blindness, is defined as having a spherical equivalent refractive error of −6.00 diopter (D) or greater, as well as an axial length of more than 26 mm.¹ The etiology of HM remains unclear because multiple factors are involved in myopia development including environmental and genetic factors.^2,3 However, early-onset high myopia (eoHM), which is characterized by preschool onset and familial clustering, is considered to be predominantly affected by genetic factors.^4,5 To date, variants in at least 16 genes have been reported to contribute to nonsyndromic Mendelian high myopia whereas variants in at least 84 genes might contribute to syndromic high myopia.⁴ Variants in these genes only contributed to a small proportion of families with eoHM based on our previous studies, about 5.6% eoHM is due to genes⁶ responsible for nonsyndromic eoHM and about 23% of eoHM is due to genes responsible for syndromic eoHM.⁷ Of these, variants in ARR3 are the most common cause of nonsyndromic eoHM. 

Of these genes, LOXL3 is of special interest because it is one of only two genes identified so far responsible for autosomal recessive eoHM (the other one being LRPAP1).⁸ LOXL3, mapping to chromosome 2p13.3, encodes a protein functioning as a copper-dependent amine oxidase.⁹ Biallelic missense variants in LOXL3 were initially identified in one family with autosomal recessive Stickler syndrome from Saudi Arabia,¹⁰ in which the two patients had eoHM and systemic signs. Subsequently, biallelic truncation variants in LOXL3 were detected in two Chinese families with autosomal recessive eoHM alone in our previous study,¹¹ which was designed as MYP28 (OMIM 619781). MYP28 with biallelic LOXL3 truncation variants was further supported by an additional study with one Saudi Arabian family.¹² Stickler syndrome caused by biallelic missense variants was further supported by an additional family from United Arab Emirates in which one patient had high myopia whereas the other only had low myopia in one eye and low hyperopia in the other eye.¹³ It would be interesting to know whether the phenotypic variability is due to the difference in variants (missense vs. truncations) or observational bias. 

In the current study, biallelic LOXL3 variants were identified in nine unrelated families with eoHM (seven new and two previously reported families), but in none of 7163 families without eoHM (P = 2.97 × 10⁻⁸), providing firm evidence to support the causative association of MYP28 with LOXL3 truncation variants. Careful reevaluation of the clinical data from the nine patients from nine unrelated families demonstrated that extreme eoHM is the only prominent and consistent feature in these patients, although typical signs for Stickler syndrome were not observed in these patients at the age of examination. Based on the time of examination results, the initial diagnosis of eoHM is considered preferable for these nine patients with biallelic LOXL3 variants. Our data suggest that LOXL3-associated MYP28 is a common form of autosomal recessive eoHM, with a frequency comparable to LRPAP1.⁸ 

Methods

Subjects and Clinical Data Collection

All probands with a variety of eye disorders and their family members were recruited in the Pediatric and Genetic Eye Clinic, Zhongshan Ophthalmic Center, Guangzhou. Before the collection of clinical data and peripheral venous blood samples of participants, an informed consent document has been obtained from all participants or their guardians, which was voluntarily signed in accordance with the tenets of the Declaration of Helsinki. Our study was approved by the institutional review board of Zhongshan Ophthalmic Center. Genomic DNA was extracted from leukocytes acquired from the peripheral venous blood as described in the previous study.¹⁴ 

A routine ophthalmic examination has been conducted on each participant. Further specific ocular and systemic examinations were performed on patients when necessary, including optic coherence tomography (OCT), fundus photography, scanning laser ophthalmoscopy, fundus autofluorescence (FAF), electroretinography (ERG), and radiography. The foveal hypoplasia grade was according to the Leicester Grading System for foveal hypoplasia.¹⁵ Additionally, the myopic fundus classification was based on a simplified classification and grading system for myopic maculopathy proposed by Ohno-Matsui et al.¹⁶ 

Variant Detection and Assessment

Whole-exome sequencing¹⁷ or targeted-exome sequencing¹⁸ was conducted on the genomic DNA of 8389 probands including 1226 with early-onset high myopia and 7163 with other eye conditions (Supplementary Fig. S1). For whole-exome sequencing, exome capture was performed with an Agilent SureSelect Human All Exon Enrichment Kit V4 (Agilent, Santa Clara, CA, USA) array. An Agilent Technologies 2100 Bioanalyzer was used to evaluate the quality of the library and an Illumina HiSeq 2000 system (Illumina, San Diego, CA, USA) was used for library sequencing with an average depth of at least 125-fold. The sequence reads were mapped to the UCSC hg19 genome using the Burrows-Wheeler Aligner software (http://bio-bwa.sourceforge.net/). Then variants were called using GATK (https://gatk.broadinstitute.org/hc.en-us) and SAMTOOLS (http://samtools.sourceforge.net/) implementing Bayesian approaches and annotated using ANNOVA (http://annovar.openbioinformatics.org/en/latest/). 

For targeted-exome sequencing, the genomic DNA was fragmented using a Bioruptor Plus (Diagenode, Liege, Belgium) to a fragment size of approximately 200 base-pair (bp). The fragments were then processed through a KAPA HTP Library Preparation Kit (Roche, Basel, Switzerland) to generate a paired-end library. Library capture was performed with a NimbleGen SeqCap EZ Choice Library SR V5 kit (Roche), and library sequencing was subsequently completed with an Illumina NextSeq550 Mild output v2 kit (150 bp paired-end) on an Illumina Nextseq550 Analyzer (Illumina, San Diego, CA, USA). Variant calling and annotation data analysis were performed using the StrandNGS software (Karnataka, India). The gene list of the target gene panel is shown in Supplementary Table S1, which has been described in our previous study.¹⁹ 

The detected variants in LOXL3 were screened by multi-steps of bioinformatics analysis. First, low sequencing quality variants with coverage of less than five were excluded. Subsequently, synonymous variants and variants in non-coding regions without impact on splicing signals, which were precomputed by human splicing finder program (https://hsf.genomnis.com/mutation/analysis), were excluded. Variants with high minor allele frequency of 0.01 or more in the gnomAD database (https://gnomad.broadinstitute.org/) were also excluded. Then the remaining variants underwent evaluation from five in silico tools: SIFT (http://sift.jcvi.org/), Polyphen-2 (http://genetics.bwh.harvard.edu/pph2/), PROVEAN (http://provean.jcvi.org/seq_submit.php), CADD (http://cadd.gs.washington.edu), and REVEL (https://sites.google.com/site/revelgenomics/). Last, variants were classified as likely pathogenic variants after the comparison analysis with the gnomAD database and Human Gene Mutation Database (HGMD). The candidate biallelic variants were validated by the Sanger sequencing following the method described in the previous study.²⁰ The primers of candidate variants in LOXL3 (NM_032603.5 for mRNA) were designed by primer 3 (http://primer3.ut.ee/), and further co-segregation analysis was performed on the available family members. 

Haplotype Analysis

Four different 5ʹ-fluorescently labeled microsatellite markers located on chromosome 2p, including D2S2368, D2S286, D2S2333, and D2S2216, were used for analyzing allelic association in samples from patients with the recurrent LOXL3 variant. These microsatellite markers were from the ABI Prism Linkage Mapping Set Version 2 (PE Applied Biosystems, Foster City, CA, USA). The PCR process has been described in the previous study.²¹ The amplified products were subsequently detected via capillary electrophoresis on the ABI 3500 Genetic Analyzer (Applied Biosystems), The genotyping analysis was performed using GeneMapper ID-X V 1.4 software. 

Statistical Analysis

All statistical analyses in this study were performed using IBM SPSS Statistics V26.0 software (Armonk, NY: IBM Corp). The χ² test or Fisher's exact test was applied to compare the frequency of biallelic LOXL3 variants among 1226 individuals with eoHM and 7163 individuals with other eye conditions in our cohort. The statistical significance was defined as P < 0.05. 

According to our recent study,⁶ candidate pathogenic variants in three genes (LRPAP1, OPN1LW, ARR3) have been suggested as causative variants for Mendelian eoHM. Additionally, LOXL3 variants also have been reported to contribute to eoHM. The keywords “LOXL3”, “LRPAP1”, “OPN1LW”, “ARR3,” and “high myopia” were searched on PubMed and Google Scholar, and references in English related to this topic were collected. Furthermore, publications linked to variants in these four genes in the HGMD database were reviewed before September 1, 2022, and collected articles related to variants responsible for eoHM. Correspondingly, we summarized the eoHM family numbers of variants in these four genes, the variant frequency, and genotype-phenotype correlations. 

Results

Molecular Analysis

After multiple bioinformatics analyses performed on the in-house exome sequencing data, biallelic LOXL3 variants were exclusively clustered in nine unrelated families with eoHM including seven new and two previously reported families, accounting for nine of 1226 probands with eoHM in the cohort, but in none of 7163 families with other eye conditions (P = 2.97 × 10⁻⁸, Fisher's exact test) (Table 1; Table 2; Supplementary Table S2a). In total, we identified seven variants in the nine families, including three missense variants (c.371G>A/p.Cys124Tyr; c.1051G>A/p.Gly351Arg; c.1669G>A/p.Glu557Lys) and four truncation variants (c.39dupG/p.Leu14Alafs*21; c.544delC/p.Leu182Cysfs*3; c.594delG/p.Gln199Lysfs*35; 1765C>T/p.Arg589*) (Table 1). All seven variants in our cohort were confirmed by Sanger sequencing (Fig. 1A). The three missense variants were predicted to be pathogenic by at least five in silico tools (Table 1). Of the seven variants, three were absent in the gnomAD database whereas four were extremely rare with allele frequency below 0.0001 (Table 1). None of the variants were found in a homozygous state in the gnomAD database. Based on the above evidence, these seven variants were classified as pathogenic or likely pathogenic variants. Interestingly, the same truncation variant (c.39dupG/p.Leu14Alafs*21) was identified in eight of nine probands. Haplotypes were constructed by genotyping data of four microsatellite markers and the c.39dupG variant (Supplementary Fig. S2). Haplotype analysis in the eight families implied that a potential founder effect might be considered between families F7 and HM293, as well as between F2 and HM407 (Supplementary Fig. S2). Conversely, the c.39dupG variant in the remaining four families was likely derived from different origin or as a recurrent event because they showed completely different haplotypes (Supplementary Fig. S2). Of the nine patients with biallelic LOXL3 variants, all had at least one truncation allele, including six with biallelic truncations (including four homozygous) and three with one truncation and one missense allele (Fig. 1A; Table 2; Supplementary Fig. S2). 

Table 1.

 

Rare Variants in LOXL3 Predicted To Be Potential Damaging

 

Table 2.

 

Clinical Information of the Probands and Affected Siblings With Biallelic LOXL3 Variants Identified in This Study (NM_032603.5)

 

Figure 1.

 

The distribution of biallelic LOXL3 variants and genetic landscape of Mendelian eoHM based on in-house cohort and previous studies. (A) The pedigrees and sequencing chromatograph of seven new families with biallelic LOXL3 variants in this study. The biallelic LOXL3 variants of patients are shown below the pedigrees, and family numbers are shown above the pedigrees. M indicates the variant site, and plus sign indicates the normal allele. The filled squares (male) and circles (female) represent the affected individuals. The black arrow points at the proband. (B) The frequency and distribution of pathogenic or likely pathogenic variants in LOXL3 in mRNA sequences are based on the current study and HGMD database. (Accession number NM_032603.5). The LOXL3 variants identified in this study are located above the sequence and in previous pieces of literature are located below the sequence. The orange-filled pattern represents LOXL3 variants reported in our previous study; the gray-filled pattern represents LOXL3 variants reported in other previous studies. The red line indicates the truncation variants, and the black line indicates the missense variants. (C) The genetic contribution of four genes (ARR3, OPN1LW, LRPAP1, and LOXL3) to eoHM. The bar plots show the counts of families with likely pathogenic variants in these four genes based on in-house data and previous studies. The orange area represents families in our cohort, and the gray area represents reported families in other previous studies.

Based on data from our cohort, the estimated prevalence of LOXL3-associated eoHM is about 7.3 × 10⁻³ (9/1226) in eoHM population and is approximately 7.3 × 10⁻⁵ in the general population if the prevalence of eoHM is estimated to be 1% in the general population (Supplementary Table S2b). To date, three loci for autosomal recessive eoHM have been mapped, of which variants in LRPAP1 and LOXL3 have been identified in MYP23 and MYP28 but not in MYP18. Taking into consideration of pathogenic and likely pathogenic variants with curation based on the large dataset in our cohort and with reference to previously published data for ARR3, OPN1LW, LOXL3, and LRPAP1,^{6,8,10–13,22–31} variants in LOXL3 appear to be the third most common cause of eoHM and potentially a common cause of autosomal recessive eoHM (Fig. 1, Supplementary Tables S3, S4). 

The Clinical Phenotypic Features

The clinical characteristics of nine patients with variants in LOXL3 are briefly presented in Table 2. The main initial symptom was nearsightedness without photophobia except for the proband F4-II:1 and the proband HM407-II:1. The main complaint of F4-II:1 was rhegmatogenous retinal detachment in the left eye due to a soccer ball hit, but the medical record showed that his spherical equivalent refraction was −20.00 diopters in both eyes before retinal detachment. Similarly, the chief complaint of HM407-II:1 was retinal detachment in the right eye due to ocular trauma, but the patient had developed high myopia in both eyes before the age of seven as previously reported.¹¹ Among these patients, the best corrected visual acuity (BCVA) at the last visit ranged from 0.04 to 0.9 (median = 0.5, decimal), and the median age at the last visit was five years (range 3.5–15 years) (Fig. 2B). Based on available optometry reports of these probands at the first visit (17 eyes of nine individuals), the average spherical equivalent was −14.99 ± 4.06 diopters (range −10.75 to −24.75 diopters), and the median age at the first visit was five years (Supplementary Table S5; Fig. 2A). According to a follow-up study of retinoscopy results, the MYP28 associated with LOXL3 showed nonprogressive myopia, characterized by a slow progression of refraction error and steady improvement of BCVA (Fig. 2). 

Figure 2.

 

The distribution and progression of spherical equivalent refractive error in the right eye and of best corrected visual acuity for patients with biallelic LOXL3 variants in our cohort. (A) The scatter plot shows the spherical-equivalent refractive error (SER) distribution and progression of five patients with biallelic LOXL3 variants. The light red dotted line represents SER of −6.00 diopters, indicating the lower limit of high myopia refraction. The red dashed curve is the best nonlinear fit of all data points, which demonstrates the slow SER progression with age. (B) The scatter plot shows the BCVA changing with the age of eight patients with biallelic LOXL3 variants. The light red dotted line represents BCVA of 0.3 decimal, indicating the legal upper limit of low vision. The red dashed line represents the best nonlinear fit of all data points, which demonstrates the slow improvement in BCVA with age.

According to the available clinical data, the fundus photography examination revealed classical myopic fundus in 17 eyes of these nine patients (Fig. 3). Based on the myopic maculopathy classification,¹⁶ tessellated fundus (Category 1) was observed in 10 eyes from five patients, and diffuse choroidal atrophy (Category 2) was observed in seven eyes from four patients. Furthermore, the myopic peripapillary crescent was detected in five eyes from three patients, posterior staphyloma was detected in one eye from one patient, and none were found to have choroidal neovascularization, radial perivascular chorioretinal degeneration, radial or circumferential lattice degeneration (Fig. 3; Table 2). The fundus autofluorescence pictures show an inconspicuous hypofluorescence spot in the foveal area. No characteristic vitreous abnormalities such as membranous or beaded vitreous anomalies were observed in these patients. The available OCT scans revealed Grade I to Grade II foveal hypoplasia in three patients (Fig. 3). The available ERG examination results demonstrated a mild reduction in cone response (Figs. 4B–D). For the extraocular features, no remarkable skeletal findings were observed, including slipped epiphysis or Legg-Perthes-like disease; scoliosis, spondylolisthesis, Scheuermann-like kyphotic deformity, or osteoarthritis before age 40. The available radiography images showed nonsignificant changes in the knee and elbow joints of two probands (F5-II:1 and F7-II:2) (Fig. 5; Supplementary Fig. S3). The available audiological evaluation was normal (from 0 to 20 dB HL across the frequency range) (Supplementary Figs. S3I–L). Additionally, these patients did not exhibit noticeable orofacial abnormalities, including flat midface, cleft palate, or micrognathia (Supplementary Figs. S3A–H). 

Figure 3.

 

The fundus images and optical coherence tomography scans of three patients in this study. (A, B) The traditional fundus images of patients (F2-II:2 and F7-II:2) showed the tessellated fundus. The OCT scans demonstrate Grade I foveal hypoplasia. (C) The traditional fundus images of the patient (F5-II:1) show the tessellated fundus and peripapillary crescent in the right eye. The OCT scans demonstrate Grade II foveal hypoplasia. The ultra-widefield fundus images of these three patients also reveal typical myopic fundus changes and no signs of radial perivascular chorioretinal degeneration, circumferential lattice degeneration, or radial lattice degeneration are observed in the right eyes. No other changes are observed except for the faint hypofluorescent spots in the fovea area in the fundus autofluorescence photographs of these three patients, which is consistent with clinical findings in high myopia but not similar to cone-rod dystrophy or retinitis pigmentosa.

Figure 4.

 

The ERG recordings of right eyes in three patients with biallelic LOXL3 variants. (A) The ERG results from a normal control eye. (B–D) The ERG findings of three patients (F1-II:1, F2-II:2, and F7-II:2) reveal the mild reduction of cone responses and normal rod responses.

Figure 5.

 

The X-ray images of two patients with likely pathogenic variants in LOXL3. (A) The X-ray films of the knee and ankle of the proband (F5-II:1) show genu valgum. (B–D) Pelvis, elbow, wrist, and ankle joint radiographs of the proband F7-II:2 reveal nonspecific changes. (E) The X-ray film of the spine of the proband F7-II:1 shows no abnormalities.

Discussion

In the current study, seven pathogenic or likely pathogenic LOXL3 variants were screened from in-house exome sequencing data of 8389 patients with different eye conditions. The biallelic LOXL3 pathogenic variants occurred exclusively in nine unrelated patients with the same phenotype (eoHM), which provided solid evidence of association of MYP28 with LOXL3 (P = 2.97 × 10⁻⁸, Fisher's exact test; P = 3.81 × 10⁻⁶, transmission disequilibrium test). The exclusion of variants in other known genes responsible for nonsyndromic and syndromic Mendelian high myopia provides additional support. All nine patients from the nine families had at least one allele with truncation variants, including six with biallelic truncations and three with one truncation allele and one missense, which are different from the two cases with biallelic missense variants and Stickler syndrome. For all nine cases in nine unrelated families from our cohort, the manifestation of eoHM is the only main and consistent feature that firmly confirms the causative roles of biallelic LOXL3 variants in the development of MYP28. 

LOXL3-associated MYP28 is likely a common form of autosomal recessive eoHM. To date, three loci (MYP18, OMIM 255500; MYP23, OMIM 615431; MYP28, OMIM 619781) have been reported for autosomal recessive eoHM, of which variants in LRPAP1 and LOXL3 have been reported to be responsible for MYP23⁸ and MYP28¹¹, respectively. So far, variants in LRPAP1 and LOXL3 have been reported in 12 families each (Fig. 1C).^{8,10–13,26,28–30} These data together with our recently reported data³² suggest that MYP28 ranks the third position for eoHM with identified variants, following the MYP26 because of the ARR3 variant and MYP1 because of the OPN1LW variant. Hence, the findings of the present study support that the LOXL3-associated MYP28 represents the common type of autosomal recessive extreme eoHM and provide clues into LOXL3-associated with MYP28 in two main ways, including molecular basis, as well as clinical characteristics, which will have important implications for the treatment and diagnosis of eoHM. 

In contrast to the clinical data observed in more than 100 cases of Stickler syndrome with variants in COL2A1 or COL11A1 in our clinic, the nine patients with biallelic LOXL3 variants only had extreme eoHM as a constant feature but without typical signs observed in Stickler syndrome, including vitreous changes or retinal abnormalities (lattice degeneration, retinal hole, retinal detachment, or retinal tear), cleft palate, characteristic facial features, high-frequency sensorineural hearing loss, skeletal findings (such as slipped epiphysis or Legg-Perthes-like disease; scoliosis, spondylolisthesis, or Scheuermann-like kyphotic deformity; osteoarthritis before age 40).^33–35 For the extraocular manifestations, no significant skeletal abnormalities, midface development, palate malformation was observed in these nine patients as described in the result. The available auditory evaluation was normal, and the X-ray images demonstrated nonsignificant changes. For the ophthalmologic presentations, except for two patients who had retinal detachment due to ocular trauma, the ultra-wide-field fundus imaging showed typical myopic fundi but without circumferential or radial lattice degeneration.^36–38 The vitreous opacity is also the key to the diagnosis of Stickler syndrome, consisting of two main subgroups: membranous and beaded vitreous anomalies.³⁹ Nevertheless, such characteristic vitreous abnormalities were absent in our patients at the age of examination. Although it is apparent that eoHM is the main finding in the nine patients described in the current study, the possibility of Stickler syndrome cannot be totally ruled out because variable phenotypes have been described for Stickler syndrome, in which atypical phenotypes may mimic eoHM in some cases as we described before.⁴⁰ However, the initial diagnosis of eoHM is preferable for the nine cases at the age of examination in the current study. 

Apart from our study (including two previously reported families), so far only five patients in three families have been reported to have biallelic variants in LOXL3 by other researchers, in which four patients in two families with biallelic missense variants were described to have Stickler syndrome (three of the four patients had eoHM)^10,13 whereas the remaining one with biallelic truncations had eoHM.¹² Except for the difference in phenotypes, however, the genotypes of the nine patients in the current study, biallelic variants with one or two truncations, are different from those reported by the two families with Stickler syndrome. Whether different genotypes contributed to different phenotypes waits to be confirmed in future studies. It has been well recognized in previous studies that missense variants may result in significantly different phenotypes from truncation variants both in dominant and recessive (even one of the two alleles) diseases, as seen in our previous studies on hereditary eye diseases with mutations in PAX6,⁴¹ CPAMD8,⁴² or RPE65.⁴³ Admittedly, some clinical features related to Stickler syndrome may occur or become more obvious in the teenage period, such as vitreous opacity, peripheral retinal lattice degeneration, and part of skeletal findings.^44,45 As seen in the nine patients described in the current study, eoHM appears to be the predominant finding at the age of examination. However, further validation of these findings would require long-term follow-up and observation of the relevant phenotypes. 

Except for the aspects mentioned above, LOXL3-associated MYP28 has been found to own unique characteristics, which is different from other common types of Mendelian eoHM such as LRPAP1-associated MYP23 and ARR3-associated MYP26. Initially, the extreme high myopia has been identified as the main clinical manifestation in LOXL3-associated MYP28 in the current study. All carriers in this study had an extremely high degree of refraction (at least −10.75 diopters) at the first visit and showed nonprogressive high myopia with age. The degree of myopic refraction in LOXL3-associated MYP28 (−14.99 ± 4.06 diopters) at the first visit is comparatively more severe than in ARR3-associated MYP26 (−12.58 ± 4.83 diopters) as previously described,⁶ but comparatively milder than in LRPAP1-associated MYP23 (−17 diopters or greater).⁸ In addition, progression of the myopia is slower in MYP28 than in MYP26.⁶ The clinical appearance of the retina in MYP28 only demonstrated typical myopic fundus changes, as well as the milder grade of myopic maculopathy, including a tessellated fundus (C1) and diffuse atrophy (C2), based on META-PM classification.¹⁶ Subsequently, as described previously, no noticeable changes were observed in FAF images of patients with LOXL3 variants in our study, except for the insignificant hypofluorescent spot in the fovea area, which were different from the FAF manifestations of cone-rod dystrophy, retinitis pigmentosa, and Stickler syndrome.^46–48 Besides, the ERG changes in these patients only demonstrated a mild reduction in cone response, which is consistent with ERG changes in high myopia^49,50 and similar to or even milder than the ERG changes in ARR3-associated MYP26.⁶ The available OCT images demonstrated Grade I to Grade II foveal hypoplasia in three patients. However, according to the regular follow-up examination results, all patients have progressive BCVA with age, and nobody has been observed to have nystagmus. Even these three patients with foveal hypoplasia have fairly good visual acuity based on the latest examination, which is consistent with the hypothesis suggested by Marmor et al.⁵¹ that the foveal pit is not the indispensable condition for the development of foveal cone specialization, either anatomically or functionally. In this study, rhegmatogenous retinal detachment caused by ocular trauma had been observed in two teenagers with LOXL3-associated MYP28, indicating that patients with MYP28 might have a high risk of retinal detachment at a young age. 

The contribution of genetic factors to high myopia has been postulated for a long time. Through whole-exome sequencing and whole-genome sequencing, rare variants in several genes including ARR3, LOXL3, LRPAP1, CCDC111, NDUFAF7, P4HA2, SCO2, UNC5D, BSG, SLC39A5, CTSH, and ZNF644, have been identified to cause myopia.^3,52 The majority of these genes are autosomal inherited, and several genes were inherited in other inheritance patterns. Based on our recent study and previous studies, LOXL3 and LRPAP1 are considered the potential causative genes for autosomal recessive high myopia.⁵² Indeed, these two genes both belong to the TGF-beta signaling pathway. LOXL3, a copper-dependent amine oxidase, has been reported to contribute to the crosslinking of collagen and elastin in the extracellular matrix (ECM), which can enhance the stiffness and stabilization of ECM.⁵³ The TGF-beta mediated both SMAD and JNK signaling pathways can induce the expression of LOXL3.^54,55 Besides, LRPAP1 encodes chaperone of lipoprotein receptor-related proteins LRP1 and the deficiency of LRP1 can perturb the TGF-beta regulation.⁸ Coincidentally, convincing evidence suggested that TGF-β expression plays an essential role in the modulation of ECM in the sclera, leading to the elongation of the ocular axis, enlargement of eye size, and the development of myopia.⁵⁶ 

In conclusion, the current study found that biallelic LOXL3 variants were exclusively enriched in nine patients with eoHM, providing firm evidence to support that causative association between biallelic LOXL3 variants and MYP28. The current study showed that LOXL3-associated MYP28 ranked third among Mendelian eoHM with identified variants, suggesting that MYP28 is a common form of autosomal recessive eoHM. Furthermore, extreme eoHM is the only constant and main feature of LOXL3-associated MYP28. The results of this study further deepen our understanding of the clinical and genetic landscape of LOXL3-associated MYP28. Moreover, identification of the genes that contribute to Mendelian eoHM, particularly in common forms, is an invaluable initial step toward understanding the molecular mechanism underlying high myopia, as well as common myopia. 

Acknowledgments

The authors are grateful to all patients and members of their families for their participation in this study. 

Supported by grants from the National Natural Science Foundation of China (81770965), the Science and Technology Planning Projects of Guangzhou (202102010271), and the Fundamental Research Funds of the State Key Laboratory of Ophthalmology. 

Disclosure: Y. Jiang, None; L. Zhou, None; Y. Wang, None; J. Ouyang, None; S. Li, None; X. Xiao, None; X. Jia, None; J. Wang, None; Z. Yi, None; W. Sun, None; X. Jiao, None; P. Wang, None; J.F. Hejtmancik, None; Q. Zhang, None