​Fifty-two affected children, 32 boys and 20 girls, were evaluated in the study. The age of the patients ranged from 4 months to 14 years, with a median age of 4.1 ± 3.1 years at the initial visit to our clinic. All patients presented with low vision, poor fixation, and nystagmus. Additional clinical findings were ocular-digital signs in 20 patients, photophobia in 14 patients, strabismus in nine patients, night blindness in four patients, and cataracts in two patients. The affected children who were able to cooperate with vision tests demonstrated reduced visual acuity, ranging from light perception to 20/100. Most of the patients were hyperopia, except two people who had myopia.

Except for the common feature of attenuated blood vessels in the retina, the appearance of the fundus in our patients ranged from a near-normal appearance to severe retinopathy. For example, our patients with the GUCY2D variant had a roughly normal appearance in their fundi, so one patient was misdiagnosed as “hyperopic, amblyopia, and nystagmus” for a long time until 14 years old. Patients with the RPGRIP1 variant exhibited a mottled appearance in their fundi. OCT scans revealed that their retinas were thinned and that the outer layers, including the myoid, ellipsoid, and outer segmental layers, had disappeared at the periphery of the retina. Patients with the CEP290 variant had a patchy appearance on the periphery of the retina. Patients with the CRB1 variant revealed a gold-foil reflection in the macular region. The OCT scan revealed a coarsely laminated retina. Patients with the LCA5 variant had a flecked appearance to their fundi. OCT scans showed that the RPE layer was atrophied and disrupted, which was supported by autofluorescence images showing fluorescence losses. Patients with variant RPE65 had a mottled fundus appearance, with one patient showing deposits of white spots in his fundus. OCT scans revealed that their retinas were thinned. One patient with the RDH12 variant showed macular atrophy and bone-like pigment deposits in his fundus. The OCT scan revealed that the retinal structure was disrupted and the entire retina was thinned (Fig. 1A, 1B, and Table 1). ERG recordings were extinguished in all patients under both photopic and scotopic conditions. The detailed clinical features are listed in Supplement Table 3.

Seventy-five variants were detected in our patients, including 29 novel and 46 previously reported variants in nine LCA-associated genes of RPGRIP1, CEP290, GUCY2D, LCA5, AIPL1, RPE65, CRB1, CRX, RDH12, and TULP1, with the highest detection rate of 26.9% (14/52) in RPGRIP1, followed by CEP290 (19.2%, 10/52), GUCY2D (15.4%, 8/52), LCA5 (9.6%, 5/52), AIPL1 (7.7%, 4/52), RPE65 (7.7%, 4/52), CRB1 (7.7%, 4/52), CRX (1.9%, 1/52), RDH12 (1.9%, 1/52), and TULP1 (1.9%, 1/52) (Supplement Fig. 2A). Thirty-eight compound heterozygous variants and 11 homozygotes variants were detected in our patients, in which 13 compound heterozygous variants were found in RPGRIP1, 10 in CEP290, six in GUCY2D, three each in AIPL1 and CRB1, two each in LCA5 and RPE65, and one in TULP1, and three homozygous variants were detected in LCA5, two of them in GUCY2D and RPE65, and one in RPGRIP1, AIPL1, CRB1, and RDH12, separately. Only one patient carried a heterozygous variant of c.571delT (p.Y191Mfs*3) in the CRX gene. All above variants included 24 missense, 15 nonsense, 27 indels, and nine splicing site variants (Supplement Fig. 2B). The variants of c.535delG (p.E179Sfs*11) in RPGRIP1 and c.421C > T (p.Q141X) in AIPL1 presented recurrently in six times and four times respectively in non-consanguineous patients. All variants and their calculated scores using the programs of SIFT, Polyphen-2, Provean, and Splice AI, as well as the ACMG classification, are listed in Supplement Table 4. Pedigrees of the families with novel variants in LCA genes are shown in Supplement Fig. 3.

The homology analysis of the proteins from five novel missense variants. (A–E) Multiple alignments of amino acids showed that leucine at position 172, 334 of GUCY2D, cysteine at position 698 of CRB1, threonine at position 1038 of CRB1, and leucine at position 58 of CEP290 were highly conserved among different species

Three-dimension model construction for novel missense variations. The red dashed lines represent hydrogen bonds. (A) The model showed that a wild-type Leucine in GUCY2D was replaced by Proline at codon 172, which would make the connective hydrogen bands lost between the Leucine and Alanine at codon 168. (B) A wild-type Alanine in GUCY2D was replaced by Proline at codon 334. (C) A wild-type polar amino acid of Cysteine in CRB1 was replaced by a nonpolar amino acid of Phenylalanine at codon 698. Different programs predicted the Gibbs free energy (ΔΔG) and showed the protein was unstable. (D) A wild-type polar amino acid of Threonine in CRB1 was replaced by a nonpolar amino acid of an amino acid of Proline at codon 1038, disrupting the hydrogen bonding. (E) A large-size wild-type Leucine was replaced by a small-size Valine at codon 58 in CEP290 ΔΔG predicting the protein became unstable. All the above-mentioned amino acid substitutions may damage the stability of the protein structure and function

The novel missense variants identified in this study exhibit a high degree of conservation among several species, including Homo sapiens, Pan troglodytes, Macaca mulatta, Canis lupus familiaris, Mus musculus, and Rattus norvegicus (Fig. 2A–E). The missense variants of c.515 T > C (p.L172P) and c.1001 T > C (p.L334P) in GUCY2D have a common feature of a wild-type aliphatic Leucine replaced by a small size Proline, which are predicted to destruct the protein stabilization due to loss of the hydrogen bonds connected Leucine with other amino acids (Fig. 3A–B). A small-size wild-type Cystine replaced by a large-size mutant-type Phenylalanine in c.2093G > T (p.C698F) of CRB1 would be predicted to destroy the structural stabilization (Fig. 3C). On the contrary, a large-size wild-type Leucine replaced by a small-size Valine in c.172C > G (p.L58V) of CEP290 would also destroy the stabilization of protein (Fig. 3E), while the missense variant of c.3112A > G (p.T1038P) in CRB1 would result in instability of the protein due to the loss of the hydrogen bonds connected Threonine with other amino acids (Fig. 3D).

Twelve cooperative children (Supplement Table 5) between the ages of 3 and 14 (averaged 7.6 ± 3.5 years) were evaluated for retinal blood supply. Additionally, a control group of 12 individuals, matched for age (mean age 7.6 ± 3.5 years) and sex (three females and nine males), were included in the study. The average diameter of retinal arteries in LCA patients was calculated to be 43.6 ± 3.8 μm, while in the normal controls, it was 51.7 ± 2.6 μm. The average diameter of retinal veins in LCA patients was measured to be 60.7 ± 4.3 μm, whereas in the normal controls, it was 62.4 ± 2.5 μm. The retinal arteries were narrowed significantly in the LCA patients than in the normal controls (P < 0.001). However, the average diameter of retinal veins was not different significantly between the patient and controls (P = 0.159) (Fig. 4).

Scatter diagram of vessel diameters of LCA patients (LCA) and controls (Ctrl). ***P < 0.001

An average peak systolic velocity (PSV) of 16.3 ± 5.4 cm/s was observed in the ophthalmic artery (OA) of the patients, while the normal controls exhibited a PSV of 23.5 ± 6.4 cm/s. The PSV in the OA was found to be significantly decreased in the LCA eyes compared to the normal eyes (P = 0.0132). The OA showed an average pulsatility (PI) value of 3.6 ± 1.3 patients and 2.5 ± 1.1 in the normal controls. The PI in the OA was slightly increased in the patients than in the controls (P = 0.0488). However, the other two parameters, EDV and RI, did not exhibit significant differences between the patients and the normal controls in the OA. In addition, all four hemodynamic parameters were not found to have a difference between the patients and the normal controls in the CRA, PCA (Table 2), CA, ICA, and ECA (Supplement Table 6).