Relative risk of retinal detachment in COL2A1 compared to COL11A1 Stickler syndrome: an individual patient data meta-analysis

Abstract

Topic:

To compare the relative risk of retinal detachment between patients with COL2A1 and COL11A1 Stickler syndrome.

Clinical Relevance:

It is unclear whether the rate of retinal detachment differs between COL2A1 and COL11A1 Stickler syndrome. Previous studies included too few patients to detect a difference between genotypes.

Methods:

Individual patient data meta-analysis of cohort studies, case-control studies, cross-sectional studies, case series, and case reports across the MEDLINE, Embase, Scopus, Web of Science Core Collection, and Web of Science Preprint Citation Index databases from 1991 to 2025. Articles providing eye examination results in subjects with genetically confirmed COL2A1 or COL11A1 Stickler syndrome were included. From the included articles, individual patient data on affected gene, age at last follow-up, and presence or absence of retinal detachment were extracted. Patients who had prophylactic retinopexy were excluded. A mixed effects logistic regression adjusted for clustering by article and by family was used to determine the relative risk of retinal detachment. Risk of bias was evaluated using the JBI Critical Appraisal Checklist for Case Series. The study was prospectively registered with PROSPERO (CRD42023428144). Overall certainty of evidence was evaluated with Grading of Recommendations Assessment, Development and Evaluation.

Results:

Of the 1420 articles screened, 179 were eligible for inclusion and 141 provided individual patient data for a total of 673 patients from 430 families. Retinal detachment was present in 229 of 491 (47%) COL2A1 Stickler syndrome patients and in 51 of 182 (28%) COL11A1 Stickler syndrome patients. The relative risk of retinal detachment was 1.78 times higher in COL2A1 compared to COL11A1 Stickler syndrome (95% CI: 1.30–2.43 p < 0.001). The certainty of evidence was moderate.

Conclusion:

Our findings indicate a higher risk of retinal detachment in COL2A1 compared to COL11A1 Stickler syndrome, which may aid clinicians in determining individualized management plans for patients with Stickler syndrome. However, due to reporting biases inherent to the case series and case reports from which we obtained data, the overall certainty of evidence was rated as moderate and our analysis was limited to providing relative risk of retinal detachment, not absolute risk.

Stickler syndrome is an inherited connective tissue disorder that presents with ocular and systemic manifestations and has an estimated incidence of 1:10,000 births.^{1, 2} Stickler syndrome can result from mutations in type II, IX, or XI collagen, the structural proteins of the vitreous gel.^1–3 The vitreous appears abnormal in Stickler syndrome, with fibrils, membranes, and areas of unusual transparency.⁴ Variants in the COL2A1 gene (Online Mendelian Inheritance in Man, OMIM^®. MIM Number: #108300), encoding type II collagen, cause more than 80% of Stickler syndrome cases.^{5, 6} Variants in the COL11A1 gene (MIM Number: #604841), encoding the α1 chain of type XI collagen, cause 10–20% of Stickler syndrome cases.^{5, 7} Variants in the genes encoding the α2 chain of type XI collagen do not affect the eye since this chain is not expressed in the vitreous.⁸ Variants in the genes encoding the three chains of type IX collagen are very rare causes of Stickler syndrome.³ Ocular manifestations of Stickler syndrome include myopia, presenile cataracts, glaucoma, vitreous abnormalities, and retinal detachment, with retinal detachment being the most likely to cause permanent visual impairment.^{9, 10} Reported rates of retinal detachment due to Stickler syndrome vary from 29–65%.^{9, 11–13} Retinal detachment in Stickler syndrome occurs at a younger age compared to typical idiopathic retinal detachments and accounts for 18% of pediatric retinal detachments.¹⁴

It is unknown whether retinal detachment risk differs between COL2A1 and COL11A1 Stickler syndromes. Our literature search found 14 studies that reported retinal detachment rates in patients of both types. Total sample sizes were low, ranging from 4 to 87 subjects, and none of the studies found a statistically significant difference in retinal detachment rates between the two groups.^{5, 7, 15–26} In 7 studies, the retinal detachment rate was non-significantly higher in the COL2A1 group, in 4 studies, the rate was non-significantly higher in the COL11A1 group, and in 3 studies, the rate was the same. These findings suggest a trend towards higher retinal detachment rates in COL2A1 Stickler syndrome but are limited by a lack of power. One systematic review found no significant difference in the risk of retinal detachment between the two types, but COL11A1 Stickler syndrome patients represented only 5% of the patients.²⁷

To compare retinal detachment rates between the two most common types of Stickler syndrome, we conducted a meta-analysis of all available cases reported in the literature that included both phenotypic and genotypic data. To maximize sample size, we collected individual patient data from studies reporting cases of COL2A1 Stickler syndrome, studies reporting cases of COL11A1 Stickler syndrome, and studies reporting cases of both types. We did not evaluate the other genes associated with Stickler syndrome due to their rarity and lack of relevance to eye disease. The aim of this meta-analysis was to examine whether retinal detachment rates differ between COL2A1 and COL11A1 Stickler syndrome, which may inform frequency of follow up and efficacy of prophylactic treatments to prevent retinal detachment.

2. Methods

2.1. Search Methods for Identifying Studies

This review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guidelines for transparency and reproducibility.²⁸ It was exempt from institutional review board approval because no data was collected through intervention or interaction with an individual and no identifiable private information was available in the data.

Five databases including MEDLINE (Ovid), Embase (Elsevier), Scopus (Elsevier), Web of Science Core Collection (Clarivate), and Web of Science Preprint Citation Index (Clarivate), were searched for articles describing patients with COL2A1 or COL11A1 Stickler syndrome using the search strategy found in Supplemental document 1. No filters were applied. The start date was chosen to be 1991, the year that mutations in COL2A1 were first linked to Stickler syndrome.²⁹ Non-English articles were translated into English for review using Google translate and/or by a native Mandarin speaker (XL). Conference abstracts without a full-text article were excluded. The latest search was conducted on May 7, 2025. References of included studies were also checked for additional studies meeting inclusion criteria.

2.2. Study Selection

Two reviewers (CMC, GRF) independently screened titles and abstracts identified by the search strategy to select full text articles for review. Two reviewers (CMC, AGM) independently reviewed the full text and determined whether to include or exclude the articles. When the two reviewers disagreed, a third reviewer (ESH) acted as the deciding vote. Reviewers utilized Covidence (Veritas Health Innovation), an online systematic reviewing platform to screen and select studies.³⁰

2.3. Eligibility Criteria for Considering Studies for this Review

Reviewers first assessed eligibility of each article based on article-level inclusion/exclusion criteria (Table 1). Eligible articles were then evaluated to select eligible individual patients based on patient-level inclusion/exclusion criteria (Table 2).

Table 1:

Article-level Inclusion and Exclusion Criteria

Inclusion Criteria	Exclusion Criteria
Original clinical study of patients with Stickler or Marshall Syndrome Study contains genetic testing results showing variants in COL2A1 or COL11A1 Study contains eye exam results	Retinal detachment and/or retinal break is an inclusion or exclusion criteria for the study 100% of patients received prophylactic treatment Study reports retinal detachment grouped together with other retinal findings

Table 2:

Patient-level Inclusion and Exclusion Criteria

Inclusion Criteria

Exclusion Criteria

Patient has a diagnosis of Stickler or Marshall syndrome as defined as either: a personal test showing a likely pathogenic or pathogenic variant or a variant that segregates with disease within their family in COL2A1 or COL11A1 a relative with a likely pathogenic or pathogenic variant or a variant that segregates with disease within their family and a personal history of at least one clinical manifestation of Stickler syndrome Information is available for the specific patient’s age at last follow-up Information is available for the specific patient’s presence or absence of retinal detachment (if detailed eye examination results are present without mention of retinal detachment, absence is assumed)

Patient received prophylactic treatment (i.e., laser, cryotherapy, or scleral buckle prior to developing a retinal detachment) in either eye Patient has a history of retinal break without detachment in either eye

We included articles describing patients with Marshall syndrome because historically, the term Marshall syndrome was used to describe patients who would now be diagnosed with COL11A1 Stickler syndrome. Marshall syndrome has a slightly different facial phenotype compared to COL2A1 Stickler syndrome.³¹ Marshall syndrome was more frequently associated with COL11A1 variants than COL2A1 variants. Use of this term has declined in modern times because: 1) there is significant overlap between the clinical manifestations of Type 2 Stickler syndrome and Marshall syndrome; 2) the same variants in COL11A1 cause both Stickler syndrome and Marshall syndrome; and 3) families exist in which family members display both syndromes.^{7, 32}

Since the focus of our study was evaluating the relative prevalence of retinal detachment, we excluded articles in which retinal detachment or break was an inclusion or exclusion criteria since these articles would not provide meaningful data on prevalence.

Checking for retinal detachment is a standard part of the eye exam. If an article reported detailed results from an eye exam but did not explicitly report whether retinal detachment was present, we assumed that retinal detachment was absent.

2.4. Data Collection and Risk of Bias Assessment

Study data was collected and managed using REDCap electronic data capture tools hosted at the Clinical & Translational Science Institute, University of Utah.³³ Two team members (CMC, AM) independently extracted data from each included article, determined which patients were eligible for inclusion (Table 2), and extracted data for each included patient. Patients with a clinical and genetic diagnosis of COL2A1 or COL11A1 Stickler syndrome or Marshall syndrome were included if there was information about their age at last follow up and presence or absence of retinal detachment. Patients who received prophylactic treatments such as laser retinopexy were excluded to better reflect natural history of the disease. Data extracted consisted of affected gene (COL2A1 vs COL11A1), sex, race, ethnicity, retinal detachment status (presence or absence), age at last follow-up, and age at retinal detachment. When missing data were identified, corresponding authors were contacted by email twice within a two-week time frame. Two team members (AGM, CYK) checked that the data sent by authors in response to our request was consistent with the article, complete, and valid. No major issues were identified. Discrepancies between team members were resolved by a third team member (ESH). Articles by the same authors and articles written by authors who frequently collaborate were carefully compared to eliminate duplicate patients. Duplicate patients were identified in most cases by matching family and patient identification codes. When necessary, sex, age, race, ethnicity, pedigrees, eye exams, and specific genetic mutations were used to identify duplicate patients.

The risk of bias of each included article was assessed independently by two team members (CMC, AGM) using the JBI Critical Appraisal Checklist for Case Series.³⁴ We did not use the 10th question about statistical analyses since our study extracted and analyzed individual patient data rather than compiling the statistical analyses of individual articles. We added an additional question about the clinic setting, since we anticipated patients seen in an eye clinic would have a higher rate of retinal detachment compared to those seen in a primary care or ear-nose-throat clinic. When discrepancies occurred, the stricter of the two answers was used as the final answer (i.e., “no” was chosen over “unclear”, which was chosen over “yes” when there was a conflict).

Two reviewers (ESH, CMC) independently evaluated the overall certainty of evidence using the Grades of Recommendation, Assessment, Development, and Evaluation (GRADE) approach for prognostic studies.^{35, 36} This instrument is used to rate the certainty of evidence by considering the quality of the included studies, as well as risk of bias, imprecision, inconsistency, indirectness, and publication bias.

2.5. Data Synthesis and Analysis

The protocol for this study was registered with the International Prospective Register of Systematic Reviews (PROSPERO) (Registration Number CRD42023428144).³⁷ In our individual patient data meta-analysis, we pre-specified that we would analyze whether the rate of retinal detachment differs between COL2A1 and COL11A1 Stickler syndrome. Individual patient data was needed because the relevant studies were underpowered and because individual ages are needed to assess time-to-event outcomes. The primary outcome variable was the presence or absence of retinal detachment. The primary predictor variable was the affected gene. The confounding variables were time at risk for retinal detachment, sex, and race/ethnicity. We tested whether the confounding variables differed significantly between the two groups (i.e. p > 0.20), and found that none did, prompting exclusion from the final model. We used a mixed effects logistic regression to test the association of retinal detachment and the affected gene while accounting for clustering of data within families and families clustered within articles. We then used marginal estimation to estimate the gene risk ratio because the risk ratio is more intuitive than the odds ratio. The risk ratio was estimated by dividing the clustering adjusted proportion of retinal detachments for the COL2A1 gene by the clustering adjusted proportion of retinal detachments for the COL11A1 gene. To assess the effects of clustering by family and by article, we used residual intraclass correlation (ICC) from the mixed effects logistic regression.

In a non-pre-specified sensitivity analysis, we performed a shared frailty cox regression accounting for time at risk of retinal detachment and accounting for the lowest level of clustering, clustering by family, as the model only allows for one level of clustering.

Time at risk of retinal detachment in Stickler syndrome begins at birth. For patients who did not have retinal detachment, time at risk was the age at last follow-up. For patients with retinal detachment, they ceased being at risk once they experienced retinal detachment. Therefore, the time at risk was age at retinal detachment. For patients lacking information about age at retinal detachment, the age at retinal detachment was assumed to be the age at last follow-up when calculating relative risk. Age at retinal detachment was compared between the two groups with a two-tailed t-test, without adjusting for clustering. This comparison was not prespecified. In a non-prespecified analysis, we used a mixed effects model accounting for clustering to calculate the percentage of patients with retinal detachment occurring at or after 5 years of age. All statistical analyses for this study were conducted using Stata/BE 17.0.³⁸

3. Results

3.1. Study Selection and Data Collection

Through our search and screen process (Figure 1), we identified 179 articles eligible for inclusion (Supplemental table 1, Supplemental document 2). We contacted authors of 100 articles via email and obtained the requested individual patient data from 18 articles. In total, 141 articles contained sufficient individual patient data (affected gene, age at last follow-up, and presence or absence of retinal detachment) for 673 patients from 430 families (Table 3). 92 articles provided information about only COL2A1 Stickler syndrome patients, 36 articles provided information about only COL11A1 Stickler syndrome patients, and 13 articles provided information about both COL2A1 and COL11A1 Stickler syndrome patients.

Figure 1.

Flow Diagram of Study Selection.

^aSome articles contained sufficient individual patient data for some patients to be included even though we did not receive additional individual patient data from authors

Table 3:

Summary of Individual Patient Data in COL2A1 and COL11A1 Stickler syndrome

	COL2A1	COL11A1	p-value
number of patients, n	491	182
number of families, n	302	128
number of articles, n	105	49
sex, n (%)
female	195 (47%)	78 (51%)	0.39
male	218 (53%)	74 (49%)
missing	78	30
patients with retinal detachment, n (%)a	229 (47%)	51 (28%)	<0.001
age at last follow-up (years), mean ± SDa	24±19	20±18	0.04
age at retinal detachment (years), mean ± SDb	16±13	18±15	0.44
time at risk for retinal detachment (years), mean ± SDc	20±17	19±18	0.45

^aThere was no missing data for retinal detachment and age at last follow-up because these were inclusion criteria

^bAge at retinal detachment was missing for 66 COL2A1 Stickler syndrome patients and 24 COL11A1 Stickler syndrome patients

^cPlease see statistical methods for definition of time at risk

3.2. Patient Characteristics

In total, we included 491 patients with COL2A1 Stickler syndrome, and 182 patients with COL11A1 Stickler syndrome (Table 3). We defined time at risk of retinal detachment as the age at last follow-up, or age at first retinal detachment when applicable. There was no significant difference in time at risk of retinal detachment between the two groups (mean ± SD was 20 ± 17 years for the COL2A1 patients, and 19 ± 18 years for the COL11A1 patients, p-value = 0.45). There were no significant differences between the two groups in sex (p = 0.39). Race was not analyzed because most articles did not follow the categories of race as defined by the National Institutes of Health of the United States (Supplemental table 2).³⁹

3.3. Retinal Detachment

Retinal detachment in at least one eye was present in 229 of 491 (47%) COL2A1 Stickler syndrome patients and in 51 of 182 (28%) COL11A1 Stickler syndrome patients (Table 3). With our mixed effects logistic regression model, we found that the risk of retinal detachment in COL2A1 Stickler syndrome was 1.78-fold higher than in COL11A1 Stickler syndrome (relative risk 1.78, 95% CI: 1.30–2.43 p < 0.001), after accounting for clustering.

Accounting for clustering was necessary since patients from the same family were more similar in retinal detachment rates than unrelated patients and patients described in the same article were more similar in retinal detachment rates than patients described in different articles. The strength of clustering was estimated by intraclass correlation from the mixed effects model. The intraclass correlation showed that article explained 25% of the variability in retinal detachment rates (95% CI: 12–44%), and the family explained 29% of the variability in retinal detachment rates (95% CI: 15–48%) after accounting for the two genes.

Given that logistic regression does not account for time at risk, as a sensitivity analysis, we also performed a shared frailty cox regression that accounts for time at risk. We found that risk of retinal detachment was higher in COL2A1 Stickler syndrome compared to COL11A1 Stickler syndrome (hazard ratio 1.80, 95% CI 1.23–2.65, p = 0.003). The hazard ratio from the cox regression (1.80) was nearly identical to the relative risk from the mixed effects logistic regression model (1.78).

We performed a secondary evaluation of age at time of retinal detachment. This information was not part of our patient inclusion criteria and was missing for 90 out of 280 patients (32%) with retinal detachment. The mean age at time of retinal detachment was similar in the COL2A1 Stickler syndrome patients (16±13 years, n=163 patients) compared to the COL11A1 Stickler syndrome patients (18±15 years, n=27 patients) (Table 3, p = 0.44). We combined data from both groups for the 183 patients with information about age at retinal detachment from 56 articles, and evaluated the distribution of age at retinal detachment to provide information about when prophylactic laser could be helpful (Figure 2). We found that 89% (95% CI: 83–95%) of retinal detachments occurred in patients age 5 or older.

Figure 2.

Distribution of age of onset of retinal detachment (in 183 total patients pooled for COL2A1 and COL11A1 Stickler syndrome). Inset, details for age less than 5 years.

3.4. Risk of Bias Assessment

The risk of bias of each included article was evaluated using the JBI Critical Appraisal Checklist for Case Series. In response to questions on the checklist, “yes” answers indicate low risk of bias. Checklist results were similar between articles providing information about COL2A1 Stickler syndrome patients and articles providing information about COL11A1 Stickler syndrome patients (Table 4). Reponses to questions 1–3 indicated low risk of bias in the diagnosis and inclusion of patients. Responses to questions 4, 5, 6, 7, and 8 indicated significant risk of bias in a few realms. Many articles lacked information about whether all eligible patients were included from a consecutive time period (questions 4–5), indicating significant risk of selection bias. Most articles lacked information about race (question 6). Many articles were not explicit when prophylactic retinopexy was not performed and when retinal detachment was not performed (questions 7 and 8). When detailed results from an eye exam were available, absence of prophylactic retinopexy and retinal detachment was inferred when they were not explicitly described. Information about clinic setting was missing for most studies, but when information was available, rates of eye clinic setting were similar between articles providing information about COL2A1 Stickler syndrome patients and articles providing information about COL11A1 Stickler syndrome patients.

Table 4:

Risk of Bias Assessment Using the JBI Critical Appraisal Checklist for Case Series

	Articles with COL2A1 patients only	Articles with COL11A1 patients only	Articles with both patient types
JBI question 1: clear inclusion criteriaa
Yes	58/67 (87%)	20/22 (91%)	7/13 (54%)
No	9/67 (13%)	2/22 (9%)	6/13 (46%)
JBI question 2: consistent diagnostic methoda
Yes	56/67 (84%)	22/22 (100%)	11/13 (85%)
No	3/67 (4%)	0/22 (0%)	1/13 (8%)
Unclear	8/67 (12%)	0/22 (0%)	1/13 (8%)
JBI question 3: used standard diagnostic criteria
Yes	85/92 (92%)	34/36 (94%)	12/13 (92%)
No	1/92 (1%)	0/36 (0%)	0/13 (0%)
Unclear	6/92 (7%)	2/36 (6%)	1/13 (8%)
JBI question 4: patients were from a consecutive time perioda
Yes	12/67 (18%)	4/22 (18%)	1/13 (8%)
No	0/67 (0%)	0/22 (0%)	1/13 (8%)
Unclear	54/67 (81%)	18/22 (82%)	10/13 (77%)
Not applicable	1/67 (1%)	0/22 (0%)	1/13 (8%)
JBI question 5: all eligible patients includeda
Yes	7/67 (10%)	3/22 (14%)	0/13 (0%)
No	3/67 (4%)	0/22 (0%)	1/13 (8%)
Unclear	56/67 (84%)	18/22 (82%)	11/13 (85%)
Not applicable	1/67 (1%)	1/22 (5%)	1/13 (8%)
JBI question 6: clear reporting of demographics
Yes	24/92 (26%)	10/36 (28%)	5/13 (38%)
No	68/92 (74%)	26/36 (72%)	8/13 (62%)
JBI question 7: explicit reporting of key clinical information for each patient
Yes	11/92 (12%)	3/36 (8%)	1/13 (8%)
No	81/92 (88%)	33/36 (92%)	12/13 (92%)
JBI question 8: explicit reporting of key outcomes for each patient
Yes	34/92 (37%)	9/36 (25%)	5/13 (38%)
No	58/92 (63%)	27/36 (75%)	8/13 (62%)
JBI question 9: sufficient details about presenting site
Yes	81/92 (88%)	33/36 (92%)	12/13 (92%)
No	11/92 (12%)	3/36 (8%)	1/13 (8%)
clinic settingb
eye	17/38 (45%)
non eye clinic	21/38 (55%)

^anot applicable to reports of a single case,

^bwhen available

We used the GRADE approach for prognostic studies to rate the overall certainly of evidence for our primary outcome, the relative risk of retinal detachment. We rated the certainty as moderate, downgrading by one point for risk of selection bias as identified by our risk of bias assessment.

4. Discussion

We conducted an individual patient data meta-analysis to test the hypothesis that the risk of retinal detachment differs between COL2A1 and COL11A1 Stickler syndrome. We obtained data for 673 patients from 141 articles, of whom 73% had COL2A1 Stickler syndrome, and 27% had COL11A1 Stickler syndrome. Retinal detachment was present in 47% of the COL2A1 Stickler syndrome patients and 28% of the COL11A1 Stickler syndrome patients. We found that the relative risk of retinal detachment was 1.78 times higher in COL2A1 compared to COL11A1 Stickler syndrome. The evidence for increased risk was graded as moderate, indicating that we are moderately confident that the rate of retinal detachment is higher in COL2A1 Stickler syndrome, but the exact relative risk may differ substantially from our findings (1.78 times higher).

Boysen et al. conducted a systematic review on ocular complications and prophylaxis of retinal detachment in Stickler syndrome and found no significant difference in the risk of retinal detachment between COL2A1 (49%, 295 of 606 patients) and COL11A1 (38%, 13 of 34 patients) Stickler syndrome.²⁷ There are several differences between our study and the study by Boysen et al. that may explain the difference in our findings. Although the total number of patients in both studies was similar, the Boysen et al. study included only 34 patients with COL11A1 Stickler syndrome whereas our study included 182 patients with COL11A1 Stickler syndrome.²⁷ In addition, we included data from 141 articles, whereas Boysen et al. included data from only 16 articles.²⁷ This may be a result of differences in exclusion criteria, since the Boysen et al. study excluded articles with less than 10 total patients or less than 2 total families.²⁷ It could also be explained by our success in obtaining individual patient data by emailing the authors. Furthermore, Boysen et al. did not evaluate the effect of two major confounders - time at risk for retinal detachment and prophylactic retinopexy.²⁷

We found a significant clustering effect between patients described in the same article accounting for 25% of the variability in retinal detachment rates. The similarity in retinal detachment rates among patients described in the same article likely reflecting significant differences in inclusion criteria and patient population source. We also found a significant clustering effect between patients from the same family, accounting for 29% of the variability in retinal detachment rates. The similarity in retinal detachment rates among patients from the same family indicates the importance of asking patients about a history of retinal detachment in family members. We found that the relative risk of retinal detachment was 1.78 times higher in COL2A1 Stickler syndrome patients after accounting for clustering by family and by article, indicating that genetic testing is useful for providing a personalized risk assessment. While prophylactic retinopexy had been previously shown to benefit patients with COL2A1 Stickler syndrome⁴⁰ and non-genotyped Stickler syndrome,^{41, 42} the effects of prophylaxis have not been studied in COL11A1 Stickler syndrome. Since the risk of retinal detachment in COL11A1 Stickler syndrome patients is significantly lower than in patients with COL2A1 Stickler syndrome, prophylaxis may be less beneficial in this patient population.

A secondary finding of our study was that the age of onset of retinal detachment was similar in COL2A1 Stickler syndrome and COL11A1 Stickler syndrome, with a mean of 16 years and 18 years, respectively. However, there was high variability in age of retinal detachment, as indicated by the standard deviation of 13 years and 15 years, respectively. We combined data from both groups and found that 89% of reported retinal detachments occurred in patients age 5 or older. This finding is relevant to the question of when to perform prophylactic laser retinopexy. If performed at age 4, it would likely be beneficial to the 89% of patients who would otherwise experience a retinal detachment at or after age 5. In contrast, prophylaxis at age 17 (the mean age of retinal detachment in the pooled group of patients of both genotypes), would occur too late to benefit the 50% of patients who already detached before this age.

One limitation of our meta-analysis is that it provides relative risk, not absolute risk. As indicated by our risk of bias assessment, there was a significant risk of selection bias in the case series and case reports from which we obtained the data. Due to selection bias, the true prevalence of retinal detachment is likely to be lower than 47% in COL2A1 Stickler syndrome and lower than 28% in COL11A1 Stickler syndrome. We found the risk of bias was similar in articles that provided data about COL2A1 Stickler syndrome patients and articles that provided data about COL11A1 Stickler syndrome Stickler syndrome patients (Table 4). Based on this assessment, we believe that biases were similar between the two groups, and a fair comparison can be made to enable calculation of relative risk.

We accounted for the most important confounders - age and family history. However, there are some confounders that we were unable to assess due to inconsistent reporting in the included articles: race, axial length, and prior intraocular surgery. The strength of association between race and retinal detachment is fairly weak, so we do not anticipate that this biased our overall outcomes.⁴³ Higher axial length is associated with retinal detachment risk in the general population.⁴⁴ One study reported higher myopia in COL11A1 Stickler syndrome,²¹ which presumably reflects higher axial length in this population. Since COL11A1 Stickler syndrome has a lower risk of retinal detachment, it must be mediated by another factor. Although prior cataract surgery is a risk factor for retinal detachment in general, in our experience, very few Stickler syndrome patients have undergone cataract surgery prior to retinal detachment.

Our study was sufficiently powered to detect a difference in retinal detachment rates by genotype, whereas previous studies did not include enough patients of both types to detect a difference. This illustrates the difficulty of obtaining sufficient sample size to study rare pediatric retinal diseases. Large datasets from multi-center studies or de-identified datasets would be highly beneficial to answering many questions of prognosis, monitoring, and management of rare genetic diseases such as Stickler syndrome.

5. Conclusion

Our individual patient data meta-analysis found an increased relative risk of retinal detachment in patients with COL2A1 Stickler syndrome compared to COL11A1 Stickler syndrome with moderate certainty of evidence. This indicates that we are moderately confident that the rate of retinal detachment is higher in COL2A1 Stickler syndrome compared to COL11A1 Stickler syndrome, but that the absolute risk may differ substantially from our findings. Genetic testing for patients with Stickler syndrome, together with age, and family history, will aid clinicians in assessing individual risk of retinal detachment, which may influence monitoring frequency and decision to perform prophylactic retinopexy. Our finding that 11% of retinal detachments in Stickler syndrome patients occur prior to age five, indicates that prophylactic retinopexy would be most beneficial before this age.