Comparative Natural History of Visual Function From Patients With Biallelic Variants in BBS1 and BBS10

Purpose The purpose of this study was to compare the natural history of visual function change in cohorts of patients affected with retinal degeneration due to biallelic variants in Bardet-Biedl syndrome genes: BBS1 and BBS10. Methods Patients were recruited from nine academic centers from six countries (Belgium, Canada, France, New Zealand, Switzerland, and the United States). Inclusion criteria were: (1) female or male patients with a clinical diagnosis of retinal dystrophy, (2) biallelic disease-causing variants in BBS1 or BBS10, and (3) measures of visual function for at least one visit. Retrospective data collected included genotypes, age, onset of symptoms, and best corrected visual acuity (VA). When possible, data on refractive error, fundus images and autofluorescence (FAF), optical coherence tomography (OCT), Goldmann kinetic perimetry (VF), electroretinography (ERG), and the systemic phenotype were collected. Results Sixty-seven individuals had variants in BBS1 (n = 38; 20 female patients and 18 male patients); or BBS10 (n = 29; 14 female patients and 15 male patients). Missense variants were the most common type of variants for patients with BBS1, whereas frameshift variants were most common for BBS10. When ERGs were recordable, rod-cone dystrophy (RCD) was observed in 82% (23/28) of patients with BBS1 and 73% (8/11) of patients with BBS10; cone-rod dystrophy (CORD) was seen in 18% of patients with BBS1 only, and cone dystrophy (COD) was only seen in 3 patients with BBS10 (27%). ERGs were nondetectable earlier in patients with BBS10 than in patients with BBS1. Similarly, VA and VF declined more rapidly in patients with BBS10 compared to patients with BBS1. Conclusions Retinal degeneration appears earlier and is more severe in BBS10 cases as compared to those with BBS1 variants. The course of change of visual function appears to relate to genetic subtypes of BBS.

B ardet-Biedl syndrome (BBS) has a broad range of clinical features that typically include severe photoreceptor degeneration often combined with truncal obesity, postaxial polydactyly, autism-like behavior, cognitive impairment, hypogonadism, and renal anomalies, among other features. [1][2][3][4][5][6] Biallelic variants have been identified in 24 BBS genes, 7,8 where BBS1 and BBS10 together are the most commonly involved. 9,10 BBS is considered a ciliopathy as the underlying genes are expressed in primary cilia. 11,12 BBS1, together with seven other BBS proteins form a protein complex named BBSome, a key regulator of the ciliary membrane proteome, important for ciliary transport. 13,14 Whereas, BBS10 protein form a complex with two other chaperonin-like proteins responsible for BBSome assembly. 15 Bardet-Biedl syndrome is phenotypically and genetically heterogeneous, and demonstrates considerable phenotypic and mechanistic overlap with other ciliopathies, such as Joubert syndrome and Senior-Løken syndrome. 1,2,4,5,[16][17][18][19][20][21][22][23][24][25] The systemic features of BBS vary among affected individuals but photoreceptor dysfunction is a constant finding. In some patients, retinal degeneration may be the only manifestation of BBS-related variants (e.g. BBS1 and C8orf37), 26,27 which was also observed for other ciliopathy genes (e.g. CEP290) causing syndromic and non-syndromic retinal degeneration. 28 Hence, genetic testing is required to identify the pathogenic variants in BBS gene to confirm BBSrelated disease.
The retinal degeneration associated with BBS is usually early and severe. The phenotypes observed include rodcone dystrophy (RCD), cone-rod dystrophy (CORD), or cone dystrophy (COD). [29][30][31][32][33][34] Central and peripheral visual function loss are most noticeable by the second or third decade of life when 73% of affected individuals become legally blind. 1,30 Studies of murine models suggest that the photoreceptor degeneration could be due to the accumulation of non-outer segment proteins in the outer segment, rather than failure of protein delivery to the outer segment. 35 It is important to identify BBS-related disease especially that a recent mice model study suggests that BBS10-related disease could be treatable, (Drack AV, et al. IOVS 2021;62: ARVO E-Abstract 1178). Lessons learned from early gene therapy studies for Leber congenital amaurosis (LCA) due to biallelic pathogenic variants in RPE65 gene highlight the importance of natural history information in patient selection and choosing the useful outcome measures to best interpret results of clinical treatment trials. 36,37 This international collaborative effort allowed collection of data from patients with retinal degeneration and biallelic variants in the two most commonly involved BBS genes; BBS1 and BBS10 (hereon referred to as patients with BBS1 and patients with BBS10). We compared the ocular phenotype of patients with BBS1 and patients with BBS10 to gain insight into the natural history of visual function loss over time.

METHODS
This retrospective study involved nine participating centers across six countries (Belgium, Canada, France, New Zealand, Switzerland, and the United States,) and was approved by the institutional ethics review board of each participating center and the procedures followed the tenets of the Declaration of Helsinki. Patients were identified through the respective internal databases. Inclusion criteria were: (1) female or male subjects with retinal degeneration, (2) biallelic pathogenic or presumed pathogenic variants in BBS1 or BBS10 genes, and (3) availability of ocular and systemic phenotype information for at least one visit. The patient's evaluation and genetic counseling were provided according to the best standard of care practice of each institution. For those countries in the European Union, the Oviedo Convention and the Treaty of Lisbon were honored. Data collected and analyzed included de-identified demographic information, ocular and medical history, systemic phenotype, DNA genetic results and parameters of the ocular phenotype (visual acuity [VA], Goldmann kinetic perimetry [VF], and electroretinography [ERG]). In selected cases, optical coherence tomography (OCT) scans were also reviewed. Full data sets were not available for all subjects for all testing parameters and some patients only had data from one visit (Table 1).

Visual Function Assessment
Data available from full field ERGs were collected from patient' charts at different ages when possible. ERGs were performed incorporating standards of The International Society for Clinical Electrophysiology of Vision (ISCEV) using Diagnosys LLC system (Canada, France, and the United States) or RETI-port system, Roland Consult (Belgium and New Zealand). 38,39 ERG results were interpreted by each principal investigator and categorized as RCD, CORD, COD, or non-detectable (ND) at first visit. The RCD pattern referred to the reduced rod and cone photoreceptor responses with predominantly reduced rod ERGs; the CORD pattern referred to reduction in cone and rod ERG responses but predominantly reduced cone responses; and the COD phenotype referred to a reduction in cone responses with preserved rod responses. ND referred to severely reduced responses not discernable from noise. Data availability: BBS1 (n = 26 for refraction, n = 35 for cataract, n = 28 for nyctalopia, and n = 25 for photophobia); BBS10-COD (n = 2 for cataract); severe BBS10 (n = 23 for refraction, n = 19 for cataract, n = 16 for nyctalopia, and n = 17 for photophobia).

Visual Acuity
The methods of VA assessment included preferential looking Teller acuity cards for preverbal children and Snellen acuity charts, or decimal (France), for older individuals. VA data for children ≤ 5 years of age were excluded due to poor reliability. All VA measurements were converted to logarithm of the minimum angle of resolution (LogMAR). Because refractive errors were not documented for all patients in some cases it was uncertain if the best correction was used. For this reason, we refer to VA and not BCVA. For the purpose of data analyses, patients who could only count fingers (CFs), perceive hand motion (HM), had only light perception (LP) or no light perception (NLP) were assigned LogMAR values of 2.6, 2.7, 2.8, and 2.9, respectively. 40,41 Acuity of each eye was measured separately but, for analyses, VA results are presented as an average of both eyes.
For patients with ≥ 5 data points, simple linear regression analyses of VA by age were performed for data from both BBS1 and BBS10 cohorts.

Refraction
Refractions were available for 62% of patients with BBS1 and 93% of patients with BBS10. Myopia was defined as a spherical equivalent < −0.5 diopters (D), hyperopia as a spherical equivalent ≥ +1.0 D and significant astigmatism as ≥ 2 D of cylinder. Results are presented as a mean spherical equivalent from both eyes.

Visual Fields
Goldmann kinetic perimetry assessments were collected at different ages for each eye. The outcome measure was the diameter across horizontal meridians for each stimulus tested (III4e and V4e). Due to symmetry, the end point was the average value from both eyes. Longitudinal and crosssection analyses of the VF to the III4e and V4e stimuli were available for some individuals and simple linear regression analyses were performed on both cohorts as described for   (29) 7/17 (41) Liver anomalies 8/32 (25) 4/20 (20) Hearing problems 8/34 (24) 5/22 (23) Diabetes 8/34 (24) 5/21 (24) Cardiac anomalies 4/34 (12) 2/21 (10) NA, not applicable. VA. A point plot graph also was performed on both cohorts with III4e data, as described below.

Data Analysis
Data were collected and summarized using descriptive measures, including means with standard deviations (SDs) and medians with ranges for continuous variables such as age, and frequencies and percentages for categorical variables, such as gender. Simple linear regression analyses were analyzed in base R and graphed in ggplot2 (https://ggplot2. tidyverse.org) in RStudio (http://www.rstudio.com/). 42,43 Genetic Analysis  44 Prediction of the pathogenicity of variants used tools publicly available. Our approach to assess the pathogenicity of variants using predictive algorithms is outlined in Table 3. The structure of BBS10 protein was modeled with Phyre2, using the cryo-EM structure of the mammalian chaperonin TRiC/CCT (PDB ID 3IYG) and the X-ray crystal structures of GroEL/GroES (PDB IDs 1SVT and 1Q3S) as templates. [45][46][47][48][49]

Demographic Information
A total of 67 patients with biallelic disease-causing variants in the BBS1 and BBS10 genes were included. Thirty-eight of all patients with BBS had presumed pathogenic variants in BBS1 (52% female patients and 48% male patients) and 29 in BBS10 (50% female patients and 50% male patients). The majority of the cohort was of Caucasian ancestry with less than 1% of Asian origin (India/Pakistan). The information about consanguinity was not reported in the patient charts. The age of patients at their first visit ranged from 2 to 49 years with mean duration of observation time of 9.7 years and mean number of 6 visits (see Table 1).

Electrophysiological Phenotype
Overall, data from ERG assessments were available for 51 patients (76%; n = 35 for BBS1, and n = 16 for BBS10  Table  1). For these 3 patients with BBS10-COD, light-adapted (LA) photopic ERGs were severely reduced at the mean age of 22.3 years (12-39 years) and rod responses were normal (Fig. 1D). In Case 39, ERGs performed at ages 34 and 39 years showed no progression.

Visual Acuity
A total of 139 VA measurements (mean 5/patient, range of 1-15) were available for BBS1 (age ranged = 5-47.8 years); and 143 VA measurements (mean 5/patient, range of 1-13) for patients with BBS10 (5-41 years). Not all measurements were available from all subjects at all visits. In the first decade, the VA profiles of patients with BBS1 and patients with BBS10 were similar, after which the rate of VA decline increased earlier in the BBS10 cohort (approximately 15 years) than in the BBS1 cohort (approximately 25 years; Fig.  2A). The greatest vision loss was documented in the late teenage years (BBS10) and early adulthood (BBS1).
A very slow progression in VA loss was observed in three patients with BBS10 with COD. At the mean age of 20 years, VA for patients with BBS10-COD and patients with RCD were 0.5 LogMAR and 1.5 LogMAR, respectively; whereas for patients with BBS1 at the mean age of 20 years, the mean VA was 0.85 LogMAR. One patient with BBS1 developed NLP vision (2.9 LogMAR) at the age of 16
Although there was variability between the different genes and mutations involved and within age groups, there was a clear inverse relationship between age and VF diameter for both III4e and V4e isopters in patients with BBS1 and in patients with BBS10-RCD (see Fig. 2B). Despite the variability, as observed for the visual acuity changes, visual field narrowing in patients with BBS10 was earlier and somewhat greater than in patients with BBS1.
In patients with BBS10 with RCD, the III4e isopter became unrecordable by the age of 25 years, whereas at the same age, patients with BBS1 had on average ≥ 20 degrees of preserved field to this isopter. Although GVFs in patients with BBS10 with RCD were not detectable to either the III4e or V4e stimulus at the age of 33, 10 of 13 (77%) of patients with BBS1 aged 30 years or older had a recordable field (range of 5-40 degrees to the III4e, IV4e, or V4e stimuli [average 20 degrees]).
At the age of 18 years GVF to V4e was similar for both cohorts, followed by a more rapid decline of GVF to V4e after age of 20 years in patients with BBS10 with RCD. The between the open (purple) and closed (magenta) forms of GroEL/GroES. The catalytic acid of chaperonins is shown as sticks, in proximity to bound ADP analog (present in solved structure). After ATP hydrolysis, the helix rotates and moves as part of the twist mechanism of facilitated substrate protein folding that the chaperonin catalyses. Ile407 is on the opposite end of the helix, and is shown as stick-like structures. (F-H) Examples of three patients with BBS10-RCD at different ages.
oldest patients with BBS1 with a recordable field was 44.9 years, whereas for patients with BBS10 it was 25 years.
Examples of the BBS10-RCD phenotype are shown in Figures 1F-H

The "Systemic" BBS Phenotype may be Very Subtle
All participants except the three cases of BBS10-COD (cases 39-41) had one to several extraocular features reminiscent of BBS. The molecular diagnosis of the three patients with BBS10-COD was from retinal degeneration gene panel testing as they were not suspected of having BBS. The availability of information about common systemic BBS features was variable (see Table 2). 18 The most prevalent extraocular features were digital anomalies (postaxial polydactyly, syndactyly, or brachydactyly), present in 96% of patients with BBS1 and 82% of patients with BBS10; followed by developmental delay, poor coordination, and kidney and liver anomalies. Because data were not available for every feature in each patient, an estimation of the frequency of each sign was impossible. Some details on part of this cohort were previously published. [2][3][4]22,50,51

Refractive Errors Were Present in 90%
Half of the 21 documented patients with BBS1 were myopic whereas the other half were hyperopic. Twelve (57%) individuals had significant astigmatism (>2 D) in combination with either myopia or hyperopia. For patients with BBS10, myopia was present in 16 (69%) of the 27 documented cases, whereas emmetropia was only seen in 3 (13%) and astigmatism was observed in about half (46%). The myopic skew in patients with BBS10-RCD compared to patients with BBS1 was statistically significant (P = 0.046). The three patients with BBS10-COD were also myopic (spherical equivalent: −8.0 D, −1.5 D, and −4.5 D, respectively).

Nyctalopia was Common by the First Decade and Cataracts by the Second Decade
Nyctalopia was an early symptom except in the three BBS10-COD cases. Only the eldest patient with BBS10-COD experienced photophobia at the age of 36 years.
Cataracts were documented at the mean age of 18.4 years in patients with BBS10-RCD (n = 18) compared to 27.4 years in patients with BBS1 (n = 24; see Table 1).

Features of Retinal Degeneration are not Specific to BBS
Retinal features in both BBS1 and BBS10-RCD cohorts showed advanced retinal degeneration with optic disc pallor, blood vessel attenuation, retinal thinning, and maculopathy, as published previously and are not specific to BBS. 2,26 OCT showed loss of structural integrity and markedly thinned outer retina (see Figs. 1F-H) and FAF had a characteristic granular pattern as previously-reported. 2,9,52 Whereas the BBS10-COD phenotype presenting with a maculopathy, as in case 39, had well-maintained retinal lamination compared to patients with BBS10-RCD of similar ages (see Fig. 1).

Genetic Analyses
Participants had confirmed biallelic variants in BBS1 or BBS10 genes, which included 11 novel variants (32%). In the 38 patients with BBS1, there was a previously documented enrichment for a common missense variant c.1169T>G, p.Met390Arg present in 31 (81.5%) patients, of whom 24 (63%) were homozygotes (see Table 2). 9,52 It is known that 80% of Caucasian patients carry this missense variant. 53 There are suggestions that this variation is a result of "hot-spot" in the gene and is the effect of multiple mutations having occurred independently at the same nucleotide. Therefore, there is a possibility that c.1169T>G, p.Met390Arg happened once, a long time ago, and was spread by emigration from its source community. 54,55 The BBS10 cohort was characterized by a total of 22 presumed pathogenic variants, showing more allelic heterogeneity than seen in the patients with BBS1. The c.271dup, p.(Cys91Leufs*5) and c.145C>T, (p.Arg49Trp) variants were by far the most common variants.
Two brothers with BBS10-COD (cases 40 and 41) were compound heterozygotes for novel variants: c.226C>T, p.(Leu76Phe) and c.181G>T, p.(Glu61*); whereas the third patient was homozygous for missense variant c.1120T>C, p.(Ile407Thr), which is rare and was seen once as heterozygous in a patient with BBS, but never observed as homozygous. Previously, the variant c.226C>T, p.(Leu76Phe) was reported as a compound heterozygote and predicted damaging, while c.181G>T, p.(Glu61*) is novel, not reported in ClinVar and gnomAD. As the phenotype was different, protein modeling was performed to further validate pathogenicity of c.1220T>C, p.(Ile407Thr) and c.181G>T, p.(Glu61*) using the mammalian chaperonin TRiC/CCT (PDB 3IYG; see Fig. 1E). 46 The models produced (Phyre2) had a confidence score of 100, with 22% sequence identity, and 41% sequence homology. This type of chaperonin, an ATPase, forms oligomeric homomultimers (double ring hexadecamers) as functional units. Because the highly conserved p.(Ile407) is present at an intersubunit interface, a mutation to threonine could alter how ATP hydrolysis induces the protein conformation changes, which is vital to chaperone function. 47 The Glu61* is predicted to cause a null protein monomer, making it impossible for the large multimeric structure to form. A completely nonfunctional protein would result. It is possible that the milder phenotype associated with the Ile407 and Leu76Phe variants related to a milder effect on BBS10 function that among the other mutations noted which were mostly null.

DISCUSSION
Although patients with BBS are known to have variable phenotypic severity, retinal degeneration is the feature always present, is relentlessly progressive and leads to legal blindness in late teenage years or young adulthood. Little was known about the natural history in molecularly characterized patients. This large genotyped cohort of patients allowed comparison of the natural history of vision loss related to the most commonly involved genes, BBS1 and BBS10, together accounting for over 40% of BBS cases. 4,6,56 Murine studies have been useful in further understanding BBS phenotypes as the genetic subtypes largely recapitulate the human phenotype. 57

What did we Learn About Visual Function in Patients With BBS-related Disease?
In the last 25 years, we learned that BBS is a genetically heterogeneous group of disorders with phenotypic and molecular overlap with other ciliopathies. Efforts to characterize the ocular phenotype of clinically defined cohorts of patients with BBS revealed important basic characteristics of BBS, although these may not reflect BBS gene-related disease subtype. 30,32,33,52,[61][62][63][64][65] For example, nyctalopia is an early symptom, photophobia is variable and manifests at different ages, and the majority of patients were reported to be myopic and develop cataract in early teenage years. The early phase of the disease can be missed as it is often sine pigmento, and the fundus changes do not have BBSspecific characteristics. 32,64 Riise 30,31 and Fulton et al. 34 evaluated visual function changes in young cohorts and reported variability in VA decline and severe loss of VF (n = 18), 64 which supports our observations, except that we see differences between patients with BBS1 and patients with BBS10. ERG recordings indicated early involvement of rod photoreceptors which is similar to our findings. 32,63,64 In our cohort, when the ERG was recordable, we report a predominant RCD phenotype (78% BBS1 and 85% BBS10), a CORD phenotype in patients with BBS1 and a COD phenotype only in patients with BBS10. Cone dystrophy was previously reported in a patient with a systemic BBS phenotype and variants in BBS6, unlike our COD cases. 29 Other cases of non-syndromic BBS-related RCD disease were previously reported, 27,66 but we report for the first time a pure cone dystrophy phenotype in only one case with hand polydactyly.
In our study, refractive errors are common in BBS and correction often benefits the patients despite the retinal degeneration; myopia was most prevalent in patients with BBS10. Comparing patients with BBS1 to patients with BBS10, there was a significant difference in changes in visual acuity and visual field, changes being more severe and earlier in patients with BBS10.
Our work supports previous studies that suggested that pathogenic variants in chaperonin-like genes (MKKS/BBS6, BBS10, and BBS12) usually lead to a more severe phenotype than those with changes affecting BBSome components, such as BBS1. 4,20,67 Strengths of our study include the large, balanced, cohort size of genotyped patients, the report of a novel COD phenotype in 3 patients with BBS10-related retinal degeneration, the longitudinal data available and that the data was captured at a wide range of ages. However, availability of the data varied at each site in part owing to the fact that no formal guidelines exist for the evaluation of these patients unlike what was recently developed for Joubert syndrome. 68 In addition, the cognition and/or behavioral characteristic of some patients would not always allow comprehensive testing. These factors together with the allelic heterogeneity precludes any mutation-phenotypic interpretation.
We believe that a prospective study would better capture uniform health parameters, but these carry time-and costrelated limitations as most inherited retinal diseases progress over years. Our multicenter retrospective approach provided valuable information in a reasonable time frame. With the developments in gene therapy in Bbs mouse-models: improving the electrophysiological responses in Bbs4-/mice, and the recent success in rescuing function in the  73 there is enthusiasm and hope to make BBS-related retinal degeneration a treatable condition.
Our retrospective study on the natural history of visual function in the largest cohort of patients with BBS1 and patients with BBS10, showed that the retinal degeneration time course of BBS10-RCD is more rapidly progressive than that of BBS1-related disease, which should be considered in the planning of treatment trials for these patients.
In summary, we have highlighted differences between the BBS1 and BBS10 phenotypes. The loss in visual function for patients with BBS10 is earlier and somewhat more severe than for patients with BBS1. The gene specific phenotypic differences are supported by data of recent murine studies also showing phenotypic differences among genetic subtypes. 60 The natural history of the BBS1 and BBS10related retinal degeneration remains somewhat incomplete as in many cases the age at which the ERG became nondetectable could not be captured and was possibly earlier than documented.