Multi-ancestry genome-wide association study of neutrophil-lymphocyte ratio and polygenic risk score development to explore causal association with diabetic retinopathy

Aravind Lathika Rajendrakumar, Anand Thakarakkattil Narayanan Nair, Mehul Kumar Chourasia, Charvi Nangia, Sundararajan Srinivasan, Venkateshan Radha, Ranjit Mohan Anjana, Moneeza K Siddiqui, Weihua Meng, Viswanathan Mohan, Colin N A Palmer
doi: https://doi.org/10.1101/2024.06.19.24309194

Abstract

Background Neutrophil–lymphocyte Ratio (NLR) is a biomarker of inflammation and was associated with diabetic retinopathy (DR) in earlier studies.

Objective To investigate the genetic loci influencing NLR and to estimate the heritability and causality of DR with the NLR polygenic risk score (PRS).

Design Genome-wide association study, conditional analysis, Fine and Gray model (FGR), Mendelian Randomization (MR)

Setting Scottish and South Indian populations drawn from population cohorts and electronic medical records.

Participants 29,317 individuals, with a considerable proportion diagnosed with diabetes.

Measurements Effect estimates from GWAS to compute PRS and causal association with DR.

Results Heritability estimates for the Scottish and Indian cohorts were 35.3% and 8.7% respectively. The top Single Nucleotide Polymorphisms (SNPs) in the multi-ancestry analysis (n=29,317) were intergenic: rs1825819 (Chr4:T/C) (Beta=-0.05, p=2.00×10-9), rs2980871 (Chr8:A/G) (Beta=0.04, p=4.64×10-8), rs2227322 (Chr17:C/G) (Beta=0.07, p=4.12×10-20) and rs4808047 (Chr19:T/C) (Beta= - 0.07, p=5.93×10-12). For the construction of best-fit PRS, we used 74,377 of 55,333,12 variants. There was a dose-response relationship between the PRS and NLR. The subhazard ratio (sHR) for NLR PRS association with DR was not statistically significant sHR=1.01 (95% CI: 0.97, 1.06, p=0.48). Null associations were observed in both cross-sectional and time-based MR analyses for PRS with DR.

Limitations A substantial proportion of the dataset was used for training the PRS algorithm. Due to trans-ancestry differences, PRS and subsequent analysis were conducted only in the Scottish cohorts.

Conclusions Multiple novel intergenic SNP associations were discovered, complementing those previously identified. Of these, some SNPs were also associated with genes known to regulate white blood cells, but not specifically NLR. More studies are required to confirm the causality between systemic inflammation and DR.

Primary Funding Source National Institute for Health Research, Pioneer and Leading Goose R&D Program of Zhejiang 2023, and the Ningbo International Collaboration Program 2023.

Introduction

Microvascular abnormalities are not observed in certain individuals despite having long-term diabetes but the exact reasons for this are ill understood.1–4 In complex diseases, the genetic liability of an individual, can be summarised into an informative risk score that can be used in disease prediction models.5,6 Diabetes and inflammation share overlapping biological mechanisms.7 The global burden of diabetic retinopathy (DR) was close to 23% and more than 6% experience vision problems.8 Increased concentrations of several inflammatory molecules were noted in tissues and vitreous fluids of individuals with DR.9 Genetic modification of specific inflammatory pathways was shown to arrest the progress of DR.10 Leukocyte recruitment in DR was found to prolong the inflammatory activity.11 Neutrophil–Lymphocyte Ratio (NLR) was shown to be more clinically informative for several clinical conditions, than a more established marker of inflammation such as C-reactive protein (CRP) in interferonopathies.12 Hence, modeling the joint variations in neutrophils and lymphocytes could be more informative than using them individually as markers of inflammation.13–17 Indeed, the NLR has been shown to be associated with diabetes and its complications.18–21 However, no studies have directly investigated the genetic causality of the observed associations. We aimed to identify new genetic loci for NLR with a strict phenotype definition. An allele score for NLR could be more powerful as it entails comprehensive information from multiple genetic loci. We hypothesized that the weighted NLR allele score could be used as an ideal instrumental variable to represent the genetic propensity for a heightened inflammatory response. We therefore performed a genome-wide association study (GWAS) for NLR and used the resulting PRS, both as an exposure and instrument variable to predict and find causal association with DR.

Methods

Cohort description and sample selection

The India-Scotland Partnership for Precision Medicine in Diabetes (INSPIRED) is a collaboration between Dr. Mohan’s Diabetes Specialties Centre (DMDSC), and the associated Madras Diabetes Research Foundation (MDRF) at Chennai in South India and the University of Dundee, Scotland.In the discovery phase, we meta-analyzed 3 Scottish cohorts: Genetics of Diabetes Audit and Research in Tayside Scotland (GoDARTS-T2D), Scottish Health Research Register (SHARE), and GoDARTS non-type 2 diabetes controls (GoDARTS controls. However, a small proportion of GoDARTS controls developed diabetes after enrollment. These Scottish cohorts consist of individuals of predominantly white European ancestry. These cohorts and the data linkage strategies used for research have been described previously.22,23 The findings were replicated in the DMDSC cohort 1 and cohort 2. The main difference between the DMDSC cohorts was the sampling strategy employed for genotyping. The participants were selected retrospectively in DMDSC cohort 1. NLR PRS was computed in SHARE using GWAS estimates from the GoDARTS T2D and controls.

Genotyping

Genotyping methods for the Scottish and Indian cohorts have been described previously.24 The DNA was extracted from the stored blood plasma. A trained bioinformatician performed the quality control using standard protocols. For all the cohorts, Haplotype Reference Consortium (HRC) was used to infer the missing genotypes. We utilized the GRCh37 assembly (Genome Research Consortium human build 37) to locate SNPs. Affymetrix 6.0, Illumina Human Omni Express, and Broad Institute platforms were used for genotyping GoDARTS. Both, DMDSC and SHARE participants were assayed using the Illumina GSA platform. Alleles with a minimum allele frequency (MAF) greater than 0.05 were considered for GWAS. Additionally, we excluded single nucleotide polymorphism (SNPs) with an info score >0.70 and Hardy-Weinberg equilibrium (HWE) <1×10-10. Principal components were computed using Plink Version 1.9.25,26 We have used STREGA guidelines to report our findings.27

Definition of NLR and DR phenotypes

The Scottish diabetic retinopathy grading scheme was used to ascertain DR status (yes/no) in the Scottish cohort.28 Details of retinal screening in Tayside were described previously.29 For the time-based Mendelian Randomization (MR) analysis, the first reported DR (any grade from R1 to R4) was considered. For non-DR individuals, the maximum follow-up date was considered as the censoring date. NLR was calculated as the absolute count of neutrophils divided by the absolute count of lymphocytes from the hematology dataset. All values of neutrophils and lymphocytes above and below 5 standard deviations (SD) were removed. The change in NLR above 2.5 from the previous reading was considered and hence excluded. For, individuals having a single NLR reading, values above 5 were excluded. NLR values after diagnosis or treatment for malignancies or within a window period of 31 days after the diagnosis of infectious diseases were discarded. We also did not retain an NLR reading within 28 days after the first reading, assuming that as a case of possible visit for a disease condition.

Ethics Approval

Tayside, Scotland UK

Ethical approval for the study was provided by the Tayside Medical Ethics Committee (REF:053/04) and the study has been carried out in accordance with the Declaration of Helsinki.

SHARE, Scotland UK

Ethical approval of the study was provided by the ethics committee SHARE East of Scotland (REF: NHS REC 13/ES/0020).

DMDSC

NIHR Global Health Research Unit on Global Diabetes Outcomes Research, Institutional Ethics Committee of Madras Diabetes Research Foundation, Chennai, India. IRB number IRB00002640. Granted 24th August 2017.

Statistical Analysis

GWAS analysis and ascertainment of independent loci

The analytical approaches are shown in Supplementary Figure 1. Continuous variables were summarized as mean/median and categorical variables as frequencies and percentages. GWAS was performed in SNPtest version 2.5.2 with 10 principal components, age, sex, and diabetes status as covariates. Genome-wide Complex Trait Analysis (GCTA) software was used to identify independent genomic loci and genomic inflation was controlled during the computation process.30 We used the cojo-joint function in the GCTA suite which uses all the SNPs to calculate joint effect size without eliminating less important SNPs from the model. GWAS summary estimates were meta-analyzed with GWAMA (Genome-Wide Association Meta-Analysis) by applying the the genomic control option (-gco) in GWAMA to address inflation and assuming an inverse variance weighted model with a p value significance cut off of 5×10-8.31

Gene enrichment, functional analysis, heritability estimation, and gene network analysis

Functional Mapping and Annotation of Genome-wide Association Studies (FUMA) was used to create the regional plot, annotation, enrichment statistics, and gene-based tests. We also took annotations from gnomAD (https://gnomad.broadinstitute.org/), UCSC browsers (https://genome.ucsc.edu/), and genecards (https://www.genecards.org/) if the annotations were not available in the FUMA. GeneMANIA was used to understand the co-expression and pathway analysis (https://genemania.org). GeneMANIA predicts weighted gene networks based on well-known genetic databases to visualize gene interactions.32 Narrow-sense heritability (h2 was estimated using LDAK software.33

PRS estimation

PRSice-2 and packages in R version 3.6.1 (The R Foundation for Statistical Computing, Vienna, Austria) were used to generate PRS and perform competing risks analyses, respectively.34–36 The Clumping and Thresholding method (C+T) in PRSice-2 was utilized for SNP thinning. GoDARTS T2D and GoDARTS controls were used as the base data and SHARE as the target data.

Competing risks regression and MR analysis

MR and competing risk analyses were performed using allele score as an instrument in the SHARE cohort. Fine and Gray analysis using the riskRegression package with deaths as a competing event adjusting for age, sex, HbA1c, BMI, and creatinine.37,38 Subsequently, The PRS from the SHARE cohort was used to test the causal relationship between NLR and DR using the one-sample Mendelian randomization (MR) method. For the Mendelian randomization analysis, we experimented with both cross-sectional and time-based models for the same DR phenotypes using R codes, and the ivtools package was used for survival MR.39

Role of the Funding Source

The funder had no role in the design, conduct, or analysis of the study or the decision to submit the manuscript for publication.

Results

Baseline characteristics of the participants

The characteristics of the Scottish discovery cohort (white Europeans, n= 21,153) are shown in Supplementary Table 1. Among these, the GoDARTS had the highest sample size. Over 96% of individuals had diabetes in the Affymetrix and Illumina sub-cohorts. The median NLR for all the Scottish cohorts was above 2. GoDARTS controls had the lowest median NLR (2.04), and only 7.7% had diabetes. Participants in the GoDARTS were older in comparison to the SHARE cohort. Except for GoDARTS controls, the proportion of males in all other cohorts was above 50%. In the DMDSC (Asian Indians = 8,164), the median NLR (1.93) and the median age was lower (less than 60 years) than the Scottish cohort (Supplementary Table 2).

Discovery Phase in white European cohorts

Meta-analysis of the 3 Scottish cohorts revealed 150 significant SNPs, most of which were in Chromosome 4 and Chromosome 17 (Supplementary Figures 2-3). Out of these, 128 were from Chromosome 17 and were in high linkage disequilibrium (LD). Only 3 SNPs were independently associated with the NLR trait, of which only one, rs3826331 in chromosome 17, and the rest, rs6841652 and rs16850400 in chromosome 4 (Table 1). Among these, rs3826331 is an intron near the neutrophil gene PSMD3, rs6841652 can be mapped either to the LINCO2513 or TBC1D1 genetic regions, and the gene corresponding to rs16850400 is yet to be determined. Of the 3 SNPs, the association signal for rs16850400 was suggestive (p=2.39 x 10-6). The estimated SNP heritability was 35.3% (SD: 0.03). The genomic inflation factor (λ) was within reasonable limits for individual λ GWAS but was slightly on the higher side for the fixed effects meta-analysis (λ =1.09).

View this table:
Table 1. SNP estimates were identified as conditionally independent using GCTA in the Scottish GWAS meta-analysis

Replication phase in Asian Indian cohort

The DMSDC GWAS plots are shown in Supplementary Figures 5 and 6. The genome-wide search for SNPs in the DMDSC cohorts found two highly correlated SNPs in Chromosome 6. Table 2 shows the details of these SNPs: rs17330192 (p=1.8 x 10-8), and rs139801819 (p=2.4 x10-8) were introns found near the non-coding region of LOC102724591 (the RNA gene). Heritability estimated in the Indian population was 8.7% (SD: 0.07). During the replication effort, a reversal of SNP effects was noted (Table 3), and rs3826331 (T/C) remained statistically significant but with a reversed effect as a trait-increasing allele (Beta =0.06, p=1.5 x 10-6). We were unable to replicate rs6841652 (p=0.06) and rs16850400 (p=0.08) in the DMDSC cohort. Arguably, the MAF was relatively common in both ancestries and therefore would not have been the reason for non-replication.

View this table:
Table 2. Estimates of SNP associations with NLR from the replication analysis in the DMDSC (n=8,164)
View this table:
Table 3. Lead SNPs identified from trans-ancestry inverse variance meta-analysis of Scottish and Indian Cohorts (n=29,317)

Fixed effects trans-ancestry meta-analysis phase

The final sample size for the FE trans-ancestry meta-analysis was 29,317, and the SNP associations are shown in Supplementary Figures 6-7. We identified two novel SNPs in Chromosome 4 and 8: rs1825819 (p=2.0×10-9) and rs2980871 (p=4.6×10-8) respectively (Table 3). The genomic inflation factor post-correction was 1.07. Regional plots show that both chromosome 17 and chromosome 19 had many SNPs in high linkage disequilibrium (Supplementary Figures 8-12). Conversely, chromosome 8 and chromosome 4 had very few highly correlated SNPs.

In the FE trans-ancestry analysis, all the lead SNPs were found to be in the non-coding regions. Four lead SNPs were found statistically to be significant, of which rs2227322 had the highest trait-increasing effect size (0.07, p=4.12×10-20). This UTR5 variant was located on chromosome 17 in the CSF3 gene. This gene was responsible for granulocyte synthesis and also cytokine-related inflammatory responses (https://www.ncbi.nlm.nih.gov/gene/1440). The second notable SNP was rs4808047 on Chromosome 19 (-0.07, p=5.93×10-12) in the EPS15L1 gene and had the lowest MAF (0.28) among the lead SNPs. SNP rs2980871 in Tribbles homolog 1 (TRIB1) was another interesting finding. In mouse models, TRIB1 expression affects hepatic lipid synthesis and glycogenesis.40 On chromosome 4, rs1825819 with a MAF of 0.38 was found to be associated with NLR (-0.05, 2.00×10-9). However, there were no known previous phenotype associations or functions identified for this specific SNP, making it a novel locus for NLR. The sentinel SNPs identified in the trans-ancestry meta-analysis generally showed a consistent direction in terms of their effects, except for the SNP rs1825819. The direction of this SNP was trait-increasing in the Broad array and reversed for other cohorts.

Figure 1 shows the significant genes in the trans-ancestry meta-analysis, in the gene-based test, CLECL1 (chromosome 12), GSDMA (chromosome 17), CSF3 (chromosome 17), PSMD3 (chromosome 17), MED24 (chromosome 17), THRA (chromosome 17), CTD-3222D19.2 (chromosome 19), EPS15L1 (chromosome 19), CALR3 (chromosome 19), CHERP (chromosome 19), CTC429P9.4 (chromosome 19), MED26 (chromosome 19), and SLC35E1 (chromosome 19) were identified. CLECL1 on chromosome 12 was another important gene that was identified. Among the 13 genes reported in the gene-based test, GSDMA, CLECL1, CALR3, and SLC35E1 did not have any functional relationships with other genes in the network analysis. No gene interaction or pathway information was available for the genes CTD-3222D19.2 and CTC429P9.

Figure 1.
Figure 1.

Manhattan plot and (left) Q-Q plot (right) corresponding to the gene set analysis from the trans-ancestry GWAS summary (n= 29,317) Note: The SNPs corresponded to 9528 protein coding genes, the genome-wide significance threshold (red dashed line in the plot) was established at P = 0.05/9528 = 5.248e-6.

The independent SNPs identified in the discovery set also remained statistically significant in the FE tran-ancestry analysis. Supplementary Figure 13 illustrates the details of the number of SNPs and genes in the important genomic loci in the meta-analysis. Tissue-wise differential gene expression analysis showed that the genes were generally upregulated in the adrenal glands, whole blood, and multiple regions of the brain (Supplementary Figure 14). However, these did not surpass the statistical significance threshold. Gene expression for nine selected significant genes from the gene-based analysis is plotted in Supplementary Figure 15. PSMD3, the major gene in our analysis, along with MED24 and KLF2, had a wider gene expression for all the tissues explored in the GTEx v8 dataset. CSF3 had an increased expression level in adipose tissue, fibroblasts, the pituitary, the prostate, the fallopian tubes, and musculoskeletal tissue. GSMDA expressions were significant only in two tissues, both relating to skin. Other genes, CHERP, KLF2, MED24, PSMD3, and THRA, showed heightened expression levels in an overlapping manner for a wide variety of tissues. Significant overlap in gene enrichment was seen in chromosomes 17 and 19 (Supplementary Figure 16). In the gene set analysis, for the 10 leading pathways, only the potassium ion transport pathway remained significant after applying the Bonferroni correction (Supplementary Table 5).

We also explored the network of genes closest to the significant SNPs in the trans-ancestry meta-analysis (Supplementary Figure 17). The network analysis of significant genes in the trans-ancestry meta-analysis revealed evidence of high co-expression (56.34%), physical interaction (25.35%), and common pathways (13.44%). For the construction of the best-fit PRS, we used 74,377 of 55,333,12 variants using the clumping and thresholding (C+T) method for SNP thinning (Supplementary Figures 18-19). The strata plot indicates a dose-response relationship between allele scores and the phenotype, although this was not discernible in the two uppermost quantiles (Figure 2). The gender-wise distribution of NLR and PRS is shown in Supplementary Figure 20.

Figure 2.
Figure 2.

Strata Plot for the NLR PRS in SHARE. Note: The plot illustrates the change in the computed PRS for NLR on the x-axis and the values of NLR on the y-axis. The blue dot represents the point estimates, and the lines represent the upper and lower 95% confidence intervals for the change in NLR corresponding to the PRS. In general, the NLR phenotype increases with the predicted PRS scores, indicating that PRS can adequately explain the individual variation of the NLR.

Competing risks analyses

In the SHARE cohort (n=3,892), the Fine and Gray model did not indicate any statistically significant association with DR (n=1,610) in both adjusted and unadjusted models (Supplementary Table 6). There were 184 competing events, and the mean follow-up time was 4.8 years. The adjusted subhazards for NLR PRS with DR were 1.01 (95% CI: 0.97, 1.06, p=0.48). Thus, PRS for NLR may not be useful for predicting the long-term manifestation of diabetic retinopathy.

Mendelian randomization analysis for DR using NLR allele scores

The regression coefficient between the phenotype and PRS was statistically significant at 0.11 (p=3.41×10-15) (Supplementary Figure 21). The PRS predicted only 1.2% of the NLR variation, which improved to 3.5% after the addition of age and sex (Supplementary Table 7). We analysed 3,561 Scottish individuals (Supplementary Figures 22-23) with data on PRS, NLR, and diabetic retinopathy for the MR analysis, of which 1,537 (43.1%) had DR. The F-statistic indicates a strong relationship between the PRS and NLR (close to 60), even after adjusting for the effects of age and sex. The results from the MR analysis are given in Figure 3. The estimates were very similar between the models. Logistic-linear and TSPS estimated a non-significant effect (n= 3,561, beta = 0.22, 95% CI: - 0.25, 0.70, p=0.35). For the Cox IV method (n=3,081, DR=1216), The mean follow-up time was 4.8 years. Again, there were no statistically relevant findings between a unit change in the PRS and hazards for DR (-0.15, 95% CI: -0.65, 0.34, p=0.55). Overall, our analysis provided no evidence of a causal association between genetically determined levels of NLR and DR.

Figure 3.
Figure 3.

MR analysis of NLR allele score with any DR. Note. Log-linear Logistic linear, TSRI two-stage residual inclusion, TSPS two-stage predictor substitution, IV Instrumental Variable, SE standard error, CI confidence interval

Discussion

Our study found a divergence in the genetic architecture of NLR between South Asians and white Europeans. South Asian populations are generally underrepresented in genetic studies of NLR, and the present results may cover this gap. In general, white Europeans had a much higher NLR relative to other ancestries.41 The heritability estimates (h2) in our work also reflect this disparity. Our heritability estimates almost matched those in the previous Dutch twins’ study.42 White blood cells are genetically determined; an example is the Duffy-Null polymorphism among the African population arising due to selection pressures.43

Remarkably, NLR was significantly correlated with multiple inflammatory markers such as CRP and IL6, but not platelet-lymphocyte ratio (PLR), implying that the variation of WBC ratios could provide biological insights.42 Overall, some important genes are associated with white blood cell-related traits, but not specifically with the NLR. Chromosome 17q21 contains many genes that exert substantial pleiotropic effects across different WBC subcomponents and is highly expressed in inflammatory disorders such as asthma and diabetes. 44,45 Given diabetes and inflammation, there was evidence of PSMD3 involvement through insulin resistance and receptor tyrosine kinase (RET) signalling, respectively.46–49 CSF3, MED24, and PSMD3 genes were reported together to influence neutrophil counts in a Bayesian meta-analysis involving multiple ancestries.50 Similarly, GSDMA, THRA, EPS15L1, CALR3, CHERP, and MED24 were also associated with multiple white blood cell types, including neutrophils and lymphocytes.45 Importantly, the rest of the significant genes; CTC429P9.4, CTD3222D19.2, and SLC35E1, do not seem to have a direct link with white blood cell physiology and are therefore novel findings that merit further mechanistic studies. Another novel NLR gene was CLECL1, which takes part in the T lymphocyte pathway but is also differentially expressed in clusters of inflammatory conditions such as diabetes mellitus and dyslipidemia.51

We replicated rs6841652 of the LINCO2513/TBC1D gene. This SNP was also associated with monocyte-lymphocyte ratio and multiple SNPs, that could be traced back to the same genes were significant for NLR.52 We were also able to detect associations with the thyroid hormone receptor gene (THRA) that were previously reported for NLR in the Korean population.53 The complex cross-talk between thyroid hormones and immune cells is vital for the proper regulation of inflammatory activity.54 Most genes from the gene-based set were related in varying degrees, as elicited in the network analysis. Moreover, the gene set analysis overwhelmingly showed affinity for the potassium ion transportation pathway.

One of the strengths of the study is that, we did not transform the NLR as opposed to previous studies 53,55 As data transformation could rasie concerns of reduced statistical power and ease of beta interpretation.56 A summarized NLR could override possible bias arising from health conditions, outliers, and other unusual deviations from the original phenotype for accurate genome-wide scanning. We further refined the SNP results in each ancestry to identify independent SNPs free from LD. To our knowledge, this is the first study that details the generation of an allelic summary score for NLR and uses it for a causal analysis for DR. We could not replicate the NLR association for DR in the observational study using the PRS.57–59 The high F statistics for the IV-phenotype relation do not completely eliminate the possibility of bias.60 It is also possible that the NLR is regulated at a more downstream level by several competing inflammatory pathways or environmental stimuli. Therefore, the current findings may not be sufficient to rule out the role of inflammation in DR. The major limitation was that a substantial proportion of the dataset was used to train the PRSice-2 algorithm, leaving a much smaller sample to investigate DR. Due to ancestry differences, PRS and subsequent analysis were conducted only in the Scottish cohorts. Finally, our one sample MR analysis is relatively less powerful compared to a two-sample MR.61,62 Additional studies are required to understand the genetic mechanisms influencing the NLR between ancestrally diverse populations and subpopulations.

Conclusion

In summary, we report multiple novel intergenic SNPs associated with the NLR in addition to those identified for WBC previously and most of the identified SNPs were in the intergenic regions. There was no evidence to suggest causality between systemic inflammation represented by NLR and DR. In the context of chronic diseases, our findings may provide insights into inflammatory pathways to develop better strategies to deal with inflammatory outcomes.

Data Availability

All data produced in the present study are available upon reasonable request to the authors.

Availability of data and materials

The Scottish and DMDSC datasets are stored electronically in a secure environment. The data is not currently available in the public domain as the participant consent restricts the public sharing of the data. Researchers will require prior approval from the HIC and DMDSC for accessing the related datasets.

Grant Support

This research was funded by the National Institute for Health Research (NIHR) (INSPIRED 16/136/102) using UK aid from the UK Government to support global health research. The views expressed in this publication are those of the author(s) and not necessarily those of the NIHR or the UK Department of Health and Social Care. Weihua Meng is supported by the Pioneer and Leading Goose R&D Program of Zhejiang 2023 (2023C04049) and the Ningbo International Collaboration Program 2023 (2023H025).

Author Contributions

C.N.A.P. and A.L.R were involved in the conceptualisation, analysis, interpretation and manuscript writing. A.T.N., M.C., and C.N. were involved in GWAS analysis and heritability estimation. SS made a substantial contribution toward ensuring data quality and manuscript revision. M.K.S, V.R., R.M.A., W.M. and V.M. revised the manuscript critically for important intellectual content and provided guidance for data analysis. All authors provided approval for manuscript publication. C.N.A.P. is the guarantor of this work, had access to the database, and takes full responsibility for the integrity of the data, decision to submit and publish the manuscript.

Competing interests

None to declare.

Supplemental Material

Supplementary File 1 – Figures

Supplementary File 2 – Tables

STREGA check list

Acknowledgments

The authors would like to thank all participants in the GoDARTS, SHARE, and DMSDC cohorts. We would also like to thank the Health Informatics Centre (HIC), University of Dundee for providing access to the Scottish Data and staff of MDRF, Chennai.

Abbreviations

BMI
Body Mass Index
CI
Confidence Interval
CIF
Cumulative Incidence Function
chr: Position
Chromosome to Position
CURES
Chennai Urban Rural Epidemiology Study
DBP
Diastolic Blood Pressure
DMDSC
Dr. Mohan’s Diabetes Specialties Clinics
DR
Diabetic Retinopathy
eGFR
Estimated Glomerular Filtration Rate
FE
Fixed Effects
FGR
Fine and Gray model
FUMA
Functional Mapping aand Annotation of Genome-wide Association Studies
GCTA
Genome-wide Complex Trait Analysis
GIF
Genomic Inflation Factor
GoDARTS
Genetics of Diabetes Audit and Research in Tayside Scotland
HbA1c
Glycated Haemoglobin
HDLc
High-density lipoprotein cholesterol
HIC
Health Informatics Centre
HRC
Haplotype Reference Consortium
HWE
Hardy-Weinberg Equilibrium
LD
Linkage Disequillibrium
LDL-C
Low-density Lipoprotein -Cholesterol
MAGMA
Multi-marker Analysis of GenoMic Annotation
MDRF
Madras Diabetes Research Foundation
NLR
Neutrophil Lymphocyte Ratio
Non-HDL-C
Non–high-density Lipoprotein Cholesterol
PLR
Platelet-lymphocyte ratio
PRS
Polygenic Risk Score
RET
Receptor Tyrosine Kinase
SBP
Systolic Blood Pressure
SD
Standard Deviation
SHARE
Scottish Health Research Register
SE
Standard Error
sHR
Sub Hazard Ratio
SNP
Single Nucleotide Polymorphisms

References

  1. 1.↵
    Klein BEK, Knudtson MD, Tsai MY, Klein R. The relation of markers of inflammation and endothelial dysfunction to the prevalence and progression of diabetic retinopathy: Wisconsin epidemiologic study of diabetic retinopathy. Arch Ophthalmol. 2009;127(9):1175–82.
    OpenUrlCrossRefPubMedWeb of Science
  2. 2.↵
    Klein BEK, Myers CE, Howard KP, Klein R. Serum lipids and proliferative diabetic retinopathy and macular edema in persons with long-term type 1 diabetes mellitus: The Wisconsin epidemiologic study of diabetic retinopathy. JAMA Ophthalmol. 2015;133(5):503– 10.
    OpenUrl
  3. 3.
    Klein R, Klein BEK, Moss SE, Cruickshanks KJ. Relationship of Hyperglycemia to the Long-term Incidence and Progression of Diabetic Retinopathy. Arch Intern Med. 1994;154(19):2169–78.
    OpenUrlCrossRefPubMedWeb of Science
  4. 4.↵
    Meng W, Shah KP, Pollack S, Toppila I, Hebert HL, McCarthy MI, et al. A genome-wide association study suggests new evidence for an association of the NADPH Oxidase 4 (NOX4) gene with severe diabetic retinopathy in type 2 diabetes. Acta Ophthalmol. 2018;96(7):e811–9.
    OpenUrlCrossRef
  5. 5.↵
    Pollack S, Igo RP, Jensen RA, Christiansen M, Li X, Cheng CY, et al. Multiethnic genome-wide association study of diabetic retinopathy using liability threshold modeling of duration of diabetes and glycemic control. Diabetes. 2019;68(2):441–56.
    OpenUrlAbstract/FREE Full Text
  6. 6.↵
    Choi SW, Mak TSH, O’Reilly PF. Tutorial: a guide to performing polygenic risk score analyses. Nat Protoc. 2020;15:2759–72.
    OpenUrlPubMed
  7. 7.↵
    Tsalamandris S, Antonopoulos AS, Oikonomou E, Papamikroulis GA, Vogiatzi G, Papaioannou S, et al. The Role of Inflammation in Diabetes: Current Concepts and Future Perspectives. Eur Cardiol. 2019;14(1):50–59.
    OpenUrlCrossRef
  8. 8.↵
    Teo ZL, Tham YC, Yu M, Chee ML, Rim TH, Cheung N, et al. Global Prevalence of Diabetic Retinopathy and Projection of Burden through 2045: Systematic Review and Meta-analysis. Ophthalmology. 2021;128(11):1580–91.
    OpenUrlCrossRefPubMed
  9. 9.↵
    Wakabayashi Y, Usui Y, Okunuki Y, Kezuka T, Takeuchi M, Goto H, et al. Correlation of vascular endothelial growth factor with chemokines in the vitreous in diabetic retinopathy. Retina. 2010;30(2):339–44.
    OpenUrlCrossRefPubMedWeb of Science
  10. 10.↵
    Tang J, Kern TS. Inflammation in diabetic retinopathy. Prog Retin Eye Res [Internet]. 2011;30(5):343–58.
    OpenUrl
  11. 11.↵
    Noda K, Nakao S, Ishida S, Ishibashi T. Leukocyte Adhesion Molecules in Diabetic Retinopathy. J Ophthalmol. 2012;2012:279037.
  12. 12.↵
    Kourilovitch M, Galarza–Maldonado C. Could a simple biomarker as neutrophil-to-lymphocyte ratio reflect complex processes orchestrated by neutrophils? J Transl Autoimmun. 2022:6;100159.
    OpenUrl
  13. 13.↵
    Anderson BO, Brown JM, Harken AH. Mechanisms of neutrophil-mediated tissue injury. J Surg Res. 1991;51(2):170–9.
    OpenUrlCrossRefPubMedWeb of Science
  14. 14.
    Liew PX, Kubes P. The Neutrophil’s Role During Health and Disease. Physiol Rev [Internet]. 2019;99(2):1223–48.
    OpenUrl
  15. 15.
    WW Zhou GY An YPC. Significant difference of neutrophil-lymphocyte ratio between colorectal cancer, adenomatous polyp and healthy people. Eur Rev Med Pharmacol Sci. 2017;21(23):5386–91.
    OpenUrl
  16. 16.
    Balta S, Celik T, Mikhailidis DP, Ozturk C, Demirkol S, Aparci M, et al. The Relation Between Atherosclerosis and the Neutrophil–Lymphocyte Ratio. Clin Appl Thromb. 2016;22(5):405–11.
    OpenUrl
  17. 17.↵
    Liu CC, Ko HJ, Liu WS, Hung CL, Hu KC, Yu LY, et al. Neutrophil-to-lymphocyte ratio as a predictive marker of metabolic syndrome. Medicine (Baltimore). 2019;98(43):e17537.
    OpenUrl
  18. 18.↵
    Ulu SM, Dogan M, Ahsen A, Altug A, Demir K, Acartürk G, et al. Neutrophil-to-Lymphocyte Ratio as a Quick and Reliable Predictive Marker to Diagnose the Severity of Diabetic Retinopathy. Diabetes Technol Ther. 2013;15(11):942–7.
    OpenUrlCrossRefPubMedWeb of Science
  19. 19.
    Abdullah M. Comparative study on neutrophil-lymphocyte ratio (NLR) among diabetic cases with and without diabetic retinopathy. IJCEO. 2021;7(1):153–6.
    OpenUrl
  20. 20.↵
    Liu J, Liu X, Li Y, Quan J, Wei S, An S, et al. The association of neutrophil to lymphocyte ratio, mean platelet volume, and platelet distribution width with diabetic retinopathy and nephropathy: A meta-analysis. Biosci Rep. 2018;38(3):BSR20180172.
  21. 21.↵
    Chittawar S, Dutta D, Qureshi Z, Surana V, Khandare S, Dubey TN. Neutrophil-lymphocyte Ratio is a Novel Reliable Predictor of Nephropathy, Retinopathy, and Coronary Artery Disease in Indians with Type-2 Diabetes. Indian J Endocrinol Metab. 2017;21(6):864–70.
    OpenUrl
  22. 22.↵
    McKinstry B, Sullivan FM, Vasishta S, Armstrong R, Hanley J, Haughney J, et al. Cohort profile: the Scottish Research register SHARE. A register of peopleinterested in research participation linked to NHS data sets. BMJ Open. 2017;7(2):e013351.
    OpenUrlAbstract/FREE Full Text
  23. 23.↵
    Hébert HL, Shepherd B, Milburn K, Veluchamy A, Meng W, Carr F, et al. Cohort profile: Genetics of Diabetes Audit and Research in Tayside Scotland (GoDARTS). Int J Epidemiol. 2018;47(2):380–381j.
    OpenUrlCrossRefPubMed
  24. 24.↵
    Srinivasan S, Liju S, Sathish N, Siddiqui MK, Anjana RM, Pearson ER, et al. Common and distinct genetic architecture of age at diagnosis of diabetes in South Indian and European populations. Diabetes Care. 2023; 46(8):1515–1523.
    OpenUrl
  25. 25.↵
    Chang CC, Chow CC, Tellier LCAM, Vattikuti S, Purcell SM, Lee JJ. Second-generation PLINK: Rising to the challenge of larger and richer datasets. Gigascience. 2015;4(1):s13742– 015–0047–8.
    OpenUrl
  26. 26.↵
    Purcell S, Chang C. PLINK 1.9 [Internet]. [cited 2023 Apr 5]. Available from: https://www.cog-genomics.org/plink/1.9/
  27. 27.↵
    Little J, Higgins JPT, Ioannidis JPA, Moher D, Gagnon F, Von Elm E, et al. STrengthening the REporting of Genetic Association Studies (STREGA)--an extension of the STROBE statement. Genet Epidemiol. 2009;33(7):581–98.
    OpenUrlCrossRefPubMedWeb of Science
  28. 28.↵
    Zachariah S, Wykes W, Yorston D. Grading diabetic retinopathy (DR) using the Scottish grading protocol. Community Eye Heal. 2015;28(92):72–3.
    OpenUrl
  29. 29.↵
    Siddiqui MK, Hall C, Cunningham SG, McCrimmon R, Morris A, Leese GP, et al. Using Data to Improve the Management of Diabetes: The Tayside Experience. Diabetes Care. 2022;45(12):2828–37.
    OpenUrl
  30. 30.↵
    Yang J, Ferreira T, Morris AP, Medland SE, Madden PAF, Heath AC, et al. Conditional and joint multiple-SNP analysis of GWAS summary statistics identifies additional variants influencing complex traits. Nat Genet. 2012;44(4):369–75.
    OpenUrlCrossRefPubMed
  31. 31.
    Mägi R, Morris AP. GWAMA: Software for genome-wide association meta-analysis. BMC Bioinformatics. 2010;11(288).
  32. 32.↵
    Warde-Farley D, Donaldson SL, Comes O, Zuberi K, Badrawi R, Chao P, et al. The GeneMANIA prediction server: biological network integration for gene prioritization and predicting gene function. Nucleic Acids Res. 2010;38((Web Server issue)):W214–20.
    OpenUrlCrossRefPubMedWeb of Science
  33. 33.↵
    Speed D, Hemani G, Johnson MR, Balding DJ. Improved Heritability Estimation from Genome-wide SNPs. Am J Hum Genet. 2012;91(6):1011–1021.
    OpenUrlCrossRefPubMed
  34. 34.↵
    Choi SW, O’Reilly PF. PRSice-2: Polygenic Risk Score software for biobank-scale data. Gigascience. 2019;8(7): giz0.
    OpenUrl
  35. 35.
    Team RC. R: A Language and Environment for Statistical Computing [Internet]. Vienna, Austria: R Foundation for Statistical Computing; 2018. Available from: http://www.r-project.org.
  36. 36.↵
    Gerds AT, Blanche P, Mortensen R, Tollenaar N, Mogensen UB, Ozenne B. Risk Regression Models and Prediction Scores for Survival Analysis with Competing Risks (riskRegression). 2020.
  37. 37.↵
    Austin PC, Lee DS, Fine JP. Introduction to the Analysis of Survival Data in the Presence of Competing Risks. Circulation. 2016;133(6):601–9.
    OpenUrlAbstract/FREE Full Text
  38. 38.↵
    Austin PC, Fine JP. Practical recommendations for reporting Fine-Gray model analyses for competing risk data. Stat Med. 2017;36(27):4391–400.
    OpenUrlCrossRefPubMed
  39. 39.↵
    Sjolander A, Martinussen T. Instrumental Variable Estimation with the R Package ivtools. Epidemiol Methods. 2019;8(1):20180024.
    OpenUrl
  40. 40.↵
    Ishizuka Y, Nakayama K, Ogawa A, Makishima S, Boonvisut S, Hirao A, et al. TRIB1 downregulates hepatic lipogenesis and glycogenesis via multiple molecular interactions. J Mol Endocrinol. 2014;52(2):145–58.
    OpenUrlAbstract/FREE Full Text
  41. 41.↵
    Azab B, Camacho-Rivera M, Taioli E. Average values and racial differences of neutrophil lymphocyte ratio among a nationally representative sample of United States subjects. PLoS One. 2014;9(11):e112361.
    OpenUrlCrossRefPubMed
  42. 42.↵
    Lin BD, Hottenga JJ, Abdellaoui A, Dolan C V, De Geus EJC, Kluft C, et al. Causes of variation in the neutrophil–lymphocyte and platelet–lymphocyte ratios: a twin-family study. Biomark Med. 2016;10(10):1061–1072.
    OpenUrl
  43. 43.↵
    Reich D, Nalls MA, Kao WHL, Akylbekova EL, Tandon A, Patterson N, et al. Reduced neutrophil count in people of African descent is due to a regulatory variant in the Duffy antigen receptor for chemokines gene. PLoS Genet. 2009;5(1):e1000360.
    OpenUrlCrossRefPubMed
  44. 44.↵
    Verlaan DJ, Berlivet S, Hunninghake GM, Madore AM, Larivière M, Moussette S, et al. Allele-specific chromatin remodeling in the ZPBP2/GSDMB/ORMDL3 locus associated with the risk of asthma and autoimmune disease. Am J Hum Genet. 2009;85(3):377–93.
    OpenUrlCrossRefPubMedWeb of Science
  45. 45.↵
    Nalls MA, Couper DJ, Tanaka T, van Rooij FJA, Chen MH, Smith A V., et al. Multiple Loci Are Associated with White Blood Cell Phenotypes. PLOS Genet. 2011;7(6):e1002113.
    OpenUrlCrossRefPubMed
  46. 46.↵
    Zheng JS, Arnett DK, Parnell LD, Lee YC, Ma Y, Smith CE, et al. Genetic variants at PSMD3 interact with dietary fat and carbohydrate to modulate insulin resistance. J Nutr. 2013;143(3):354–61.
    OpenUrlAbstract/FREE Full Text
  47. 47.
    Colaprico A, Olsen C, Bailey MH, Odom GJ, Terkelsen T, Silva TC, et al. Interpreting pathways to discover cancer driver genes with Moonlight. Nat Commun. 2020;11(1):69.
    OpenUrl
  48. 48.
    Szilveszter KP, Németh T, Mócsai A. Tyrosine Kinases in Autoimmune and Inflammatory Skin Diseases. Front Immunol. 2019;10:1862.
    OpenUrl
  49. 49.↵
    Zheng JS, Arnett DK, Parnell LD, Lee YC, Ma Y, Smith CE, et al. Genetic variants at PSMD3 interact with dietary fat and carbohydrate to modulate insulin resistance. J Nutr. 2013;143(3):354–61.
    OpenUrlAbstract/FREE Full Text
  50. 50.↵
    Keller MF, Reiner AP, Okada Y, van Rooij FJA, Johnson AD, Chen MH, et al. Trans-ethnic meta-analysis of white blood cell phenotypes. Hum Mol Genet. 2014;23(25):6944–60.
    OpenUrlCrossRefPubMed
  51. 51.↵
    Veroneze R, Corbi SCT, Da Silva BR, De Rocha CS, Maurer-Morelli C V., Orrico SRP, et al. Using association rule mining to jointly detect clinical features and differentially expressed genes related to chronic inflammatory diseases. PLoS One. 2020;15(10):e0240269.
    OpenUrl
  52. 52.↵
    Kachuri L, Jeon S, DeWan AT, Metayer C, Ma X, Witte JS, et al. Genetic determinants of blood-cell traits influence susceptibility to childhood acute lymphoblastic leukemia. Am J Hum Genet. 2021;108(10):1823–35.
    OpenUrlCrossRef
  53. 53.↵
    Park B, Choe E, Kang H, Shin E, Lee S, Won S. Genetic Polymorphisms Associated with the Neutrophil–Lymphocyte Ratio and Their Clinical Implications for Metabolic Risk Factors. J Clin Med. 2018;7(8):204.
    OpenUrl
  54. 54.↵
    Perrotta C, Buldorini M, Assi E, Cazzato D, De Palma C, Clementi E, et al. The Thyroid Hormone Triiodothyronine Controls Macrophage Maturation and Functions: Protective Role during Inflammation. Am J Pathol. 2014;184(1):230–47.
    OpenUrlPubMed
  55. 55.↵
    Lin BD, Carnero-Montoro E, Bell JT, Boomsma DI, De Geus EJ, Jansen R, et al. SNP heritability and effects of genetic variants for neutrophil-to-lymphocyte and platelet-to-lymphocyte ratio. J Hum Genet. 2017;62(11):979–88.
    OpenUrlCrossRef
  56. 56.↵
    Shen X, Rönnegård L. Issues with data transformation in genome-wide association studies for phenotypic variability. F1000Res. 2013;2:200.
    OpenUrl
  57. 57.↵
    Ulu SM, Dogan M, Ahsen A, Altug A, Demir K, Acartürk G, et al. Neutrophil-to-Lymphocyte Ratio as a Quick and Reliable Predictive Marker to Diagnose the Severity of Diabetic Retinopathy. Diabetes Technol Ther. 2013;15(11):942–7.
    OpenUrlCrossRefPubMedWeb of Science
  58. 58.
    Binet F, Cagnone G, Crespo-Garcia S, Hata M, Neault M, Dejda A, et al. Neutrophil extracellular traps target senescent vasculature for tissue remodeling in retinopathy. Science. 2020;369(6506):eaay5356.
    OpenUrlAbstract/FREE Full Text
  59. 59.↵
    He X, Qi S, Zhang X, Pan J. The relationship between the neutrophil-to-lymphocyte ratio and diabetic retinopathy in adults from the United States: results from the National Health and nutrition examination survey. BMC ophthalmol. 2022;22(1):346.
    OpenUrl
  60. 60.↵
    Burgess S, Thompson SG, Chd C, Collaboration G. Avoiding bias from weak instruments in Mendelian randomization studies. Int J Epidemiol. 2011;40(3):755–64.
    OpenUrlCrossRefPubMedWeb of Science
  61. 61.↵
    Burgess S, Scott RA, Timpson NJ, Smith GD, Thompson SG. Using published data in Mendelian randomization: A blueprint for efficient identification of causal risk factors. Eur J Epidemiol. 2015;30(7):543–52.
    OpenUrlCrossRefPubMed
  62. 62.↵
    Burgess S, Timpson NJ, Ebrahim S, Smith GD. Mendelian randomization: where are we now and where are we going? Int J Epidemiol. 2015;44(2):379–88.
    OpenUrlCrossRefPubMed
Back to top
Posted June 20, 2024.
Download PDF

Supplementary Material

Data/Code
Email
Multi-ancestry genome-wide association study of neutrophil-lymphocyte ratio and polygenic risk score development to explore causal association with diabetic retinopathy
Aravind Lathika Rajendrakumar, Anand Thakarakkattil Narayanan Nair, Mehul Kumar Chourasia, Charvi Nangia, Sundararajan Srinivasan, Venkateshan Radha, Ranjit Mohan Anjana, Moneeza K Siddiqui, Weihua Meng, Viswanathan Mohan, Colin N A Palmer
medRxiv 2024.06.19.24309194; doi: https://doi.org/10.1101/2024.06.19.24309194
Twitter logo Facebook logo LinkedIn logo Mendeley logo
Citation Tools

Subject Area