Diversity in Polygenic Risk of Primary Open-Angle Glaucoma

Round 1

Reviewer 1 Report

The authors review GWAS studies identifying individual risk alleles for primary open angle glaucoma and their aggregation into genetic risk scores and polygenic risk scores. It is not clear to what extent these scores are currently used to screen for risk of glaucoma or predict worse glaucoma outcomes clinically. If that is the ultimate aim, an idea as to when that might be implemented would be useful. Several studies are cited in which the scores do correlate with measures of glaucoma severity, notably in the European ancestry patients, who were the focus of most of the GWAS studies, and less efficiently in patients of African ancestry, which is the point of the article.

 

Studies specifically in African populations are cited in which novel risk alleles were identified in addition to overlapping alleles already found in European populations. In the GIGA study, a risk allele in one African group was not found in other continental African populations, which is concerning as the number of regional populations that may need to be studied world-wide may be large. (GnomAD includes only a Yoruban African population in its 8 aggregated databases.) Additional risk alleles were found in the ADAGES study; a polygenic risk score performed well when tested in the same population from which it was derived and in which the allele would have been expected to have been enriched.

 

It is not clear whether the problem defining genetic risk scores that work well in African populations relates to lack of data (genomes) or to an informatics issue of designing a good risk tool. It is possible that there is a biological problem in that alleles that are harmful in one population are protective in another and that simply having any risk allele associated with any ancestry is not sufficient to create a good risk tool. These are speculations offered by a non-expert reader who is left with significant questions after reading the article as to why these tools cannot be redesigned statistically without having to query ancestral diversity before knowing which tool to apply. That may be necessary if genomes are so vastly different (unlikely), but it seems that what would most benefit medicine is having the test that ends up in the hands of the clinician be purely genetic and unrelated to ancestry. The authors mention optic nerve OCT as an example of a clinical test in which diversity interferes with test interpretation, but separate nomograms for Black and white patients can hardly be a good solution. The issue is probably not just informatics but the need for more data so that risk alleles are not missed – more GWAS studies in patients of African ancestry and more GWAS studies in diverse populations everywhere. Once novel risk alleles are detected in regional populations, they should be able to be included in a global polygenic risk score, although, again biology may have something to say about that simple plan.   

 

The authors’ comments in “Global Perspectives” provide essential context for the importance of glaucoma care and glaucoma research in high-risk populations, specifically those of African descent.  

 

If aspects of this review indicate to the author that I have misinterpreted their comments, they may wish to revise with shorter and simpler sentences.  

Author Response

Point 1: The authors review GWAS studies identifying individual risk alleles for primary open angle glaucoma and their aggregation into genetic risk scores and polygenic risk scores. It is not clear to what extent these scores are currently used to screen for risk of glaucoma or predict worse glaucoma outcomes clinically. If that is the ultimate aim, an idea as to when that might be implemented would be useful.

Response 1: We thank the Reviewer for identifying the need to clarify the extent to which genetic and polygenic risk scores are currently used to screen for risk of glaucoma and/or predict worse glaucoma outcomes clinically.

Recent studies and reviews outline their eventual hope for a clinically-useful risk model using GRS and PRS. The first step often cited in these works is the discovery of more POAG genetic risk factors. A review done by Choquet and colleagues note the difficulty in using SNPs obtained via GWAS as the sole clinical, predictive tool, as it does not identify causal variants. Additionally, while individual risk screening can be done via DNA sequencing, the diagnostic yield is low (20%). Generalized genetic screening for POAG may be useful for those over the age of 50, as recent studies have shown a ROC score of 0.76. However, barriers to this type of genetic testing do exist: (i) individual variants often have a small effect on disease risk compared to sociodemographic or environmental factors, (ii) the sensitivity and specificity of these tests do not bode well for cost-effectiveness, and (iii) each proposed test needs additional validation in a clinical setting.

Souzeau and colleagues began preliminary analysis of the clinical utility of screening for the MYOC disease-causing variants in POAG with positive results but are careful to note potential biases in their study design. However, the MYOC genetic variant is one of the few that results in glaucoma diagnosis early in life and with a significant effect alone. Most genetic variants that influence glaucoma arise from combined factors or interactions that would have negligible effect if presented singularly. Once again, the genetic diversity in this small sample size is lacking, making it difficult to generalize to other ancestral groups. While researchers have been making great strides in expanding the genetic and ancestral diversity in new GWAS, an exact timeline for a predictive tool cannot be established.

Additionally, Siggs and colleagues also used previously established, European-based databases such as PROGRESSA and QSkin to evaluate the effect of genetic risk on visual field worsening in POAG. Those with the highest risk scores (top 5%) progressed more quickly than the bottom 95%. However, the retrospectively evaluated cohort was not ancestrally diverse, and the researchers clarified that their findings were most likely not generalizable across more diverse populations.

 

We have added text to the “Genetic and Polygenic Risk Scores” section of the paper to further clarify these points.

 

Point 2: Studies specifically in African populations are cited in which novel risk alleles were identified in addition to overlapping alleles already found in European populations. In the GIGA study, a risk allele in one African group was not found in other continental African populations, which is concerning as the number of regional populations that may need to be studied world-wide may be large. (GnomAD includes only a Yoruban African population in its 8 aggregated databases.) Additional risk alleles were found in the ADAGES study; a polygenic risk score performed well when tested in the same population from which it was derived and in which the allele would have been expected to have been enriched.

 

Response 2: While this portion of the Reviewer’s comments does not necessarily provide a specific critique to the manuscript itself, but rather the state of genetic research and databases as a whole, we felt it necessary to include more context surrounding the overall lack of genetic data in those of African descent, even outside the scope of ophthalmology. We have added text to the paper to further clarify these points.

 

Relevant literature describes African populations as inherently diverse due to the robust evolutionary and migration patterns of the continent. Gomez and colleagues described the genetic diversity of people of African descent, though quite broad, to be highly correlated with geographic location and language. However, they noted the lack of sampling of African populations due to the current high costs of whole genome sequencing. Barriers to resources ultimately contribute to the existing disparity of genetic information when comparing the availability of information from European and African populations. This lack of genetic information exists for those of African ancestry as a whole, outside of the realm of ophthalmology. By drawing attention to the disparity on this level, we hope to communicate the fundamental need for genomic data in these underrepresented, high-risk populations.

 

Point 3: It is not clear whether the problem defining genetic risk scores that work well in African populations relates to lack of data (genomes) or to an informatics issue of designing a good risk tool. It is possible that there is a biological problem in that alleles that are harmful in one population are protective in another and that simply having any risk allele associated with any ancestry is not sufficient to create a good risk tool. These are speculations offered by a non-expert reader who is left with significant questions after reading the article as to why these tools cannot be redesigned statistically without having to query ancestral diversity before knowing which tool to apply. That may be necessary if genomes are so vastly different (unlikely), but it seems that what would most benefit medicine is having the test that ends up in the hands of the clinician be purely genetic and unrelated to ancestry.

Response 3: We thank the Reviewer for highlighting this issue of clarity in our manuscript. We have thoughtfully worked to streamline in response to these comments.

 

As of 2021, Neustaeter and colleagues note that using GRS as a screening tool in a clinical setting has not yet been validated. Their work attempts to create a GRS screening tool for glaucoma to test feasibility. This GRS was created using previously discovered SNPs in European populations only. The researchers gathered all available data on glaucoma-related SNPs and used the effect sizes listed by the Choquet meta-GWAS database. The Choquet database utilizes the UK Biobank and GERA cohorts. Of the approximately 78,000 UK Biobank participants, only 8.5% were of African descent. Even less in the ~103,000 subject GERA cohort self-reported African American, African, or Afro-Caribbean ancestry (3.3%). However, Neustaeter has not yet published their findings that would use this risk-screening tool in a prospective cohort. Similarly, most other literature that mentions using GRS has thus used it retrospectively, thus automatically ruling out possible applications as a screening tool.

Liu and colleagues underscored the need for diverse representation in GWAS. These researchers attempted to assess previously POAG-associated SNPs in those of African ancestry. These included 57 SNPs in the following genomic loci: CDKN2B-AS1, TMCO1, CAV1/CAV2, chromosome 8q22 intergenic region, and SIX1/SIX6. Unsurprisingly, these SNPs were not as significantly associated with POAG or POAG risk in the African cohort. Creating a GRS or PRS that is based on the majority of existing literature and research would yield a risk screening tool that would not be beneficial to populations not adequately represented in the current GWAS database.

Current scientific tools or risk screening initiatives cannot be tailored or “redesigned” to minority or African-descent populations specifically because these genetic risk factors have not yet been discovered, much less validated in a clinical setting.

Point 4: The authors mention optic nerve OCT as an example of a clinical test in which diversity interferes with test interpretation, but separate nomograms for Black and white patients can hardly be a good solution. The issue is probably not just informatics but the need for more data so that risk alleles are not missed – more GWAS studies in patients of African ancestry and more GWAS studies in diverse populations everywhere. Once novel risk alleles are detected in regional populations, they should be able to be included in a global polygenic risk score, although, again biology may have something to say about that simple plan.  

Response 4: We appreciate the Reviewer’s acknowledgement of the need for more GWAS in diverse populations, as that is a major point of this manuscript.

 

Point 5: The authors’ comments in “Global Perspectives” provide essential context for the importance of glaucoma care and glaucoma research in high-risk populations, specifically those of African descent. 

Response 5: We appreciate the Reviewer’s acknowledgement of the importance of this topic.

 

Point 6: If aspects of this review indicate to the author that I have misinterpreted their comments, they may wish to revise with shorter and simpler sentences. 

Response 6: We thank the Reviewer for this insight. We have edited the manuscript for clarity, removed typos, and appropriately trimmed lengthy sentences.

Author Response File: Author Response.docx

Reviewer 2 Report

Bailey et al. realized a very interesting review describing the “Diversity in Polygenic Risk of Primary Open-Angle Glaucoma”. I consider the manuscript very interesting but, at the same time, I suggest several revisions needed to improve the reliability and the completeness of the paper: 

·      The “Introduction” sections should be more detailed and improved. I suggest adding data related to the involvement of different variants in the etiopathogenesis of ocular diseases, especially with vascular components. The recent PMID: 32877751, PMID: 30523548 and PMID: 36290689 could represent a substrate able to enforce the role of considered cellular mechanisms.

 

·      Finally, manuscript requires important English revisions and typos correction.

Author Response

Point 1: The “Introduction” sections should be more detailed and improved. I suggest adding data related to the involvement of different variants in the etiopathogenesis of ocular diseases, especially with vascular components. The recent PMID: 32877751, PMID: 30523548 and PMID: 36290689 could represent a substrate able to enforce the role of considered cellular mechanisms.

Response 1: While we appreciate the Reviewer’s attention to the broad field of glaucoma biology, we respectfully chose to keep the manuscript succinct and focused on the topic of diversity in POAG genetics.

 

Point 2: Finally, manuscript requires important English revisions and typos correction.

Response 2: Such edits have been made.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

The authors addressed all suggested points.