Association of ICAM-1 Gene Polymorphisms with Diabetic Retinopathy in T2DM Patients from Northern India: Case-control and meta-analysis

Abstract

BACKGROUND

Diabetic retinopathy (DR) is a leading cause of vision loss in working-age adults, with prevalence varying by population and reaching ~ 34% in northern India. DR arises from chronic hyperglycemia–driven oxidative stress, inflammation, and microvascular dysfunction. Intercellular adhesion molecule-1 (ICAM-1) is central to leukocyte adhesion and retinal vascular injury; circulating ICAM-1 is elevated in patients and experimental models. Genetic variants in ICAM-1, notably c.1405A>G (rs5498) and c.721G>A (rs1799969), have been examined as risk markers for microvascular complications. Yet associations with DR are inconsistent across ethnicities, and robust data from northern India are limited, underscoring the need for population-specific studies.

AIM

To determine the association of ICAM-1 gene polymorphisms with DR in patients with type 2 diabetes mellitus (T2DM) from northern India.

METHODS

The present study included 614 participants: 302 patients with T2DM and DR and 312 patients with T2DM but without DR. The ICAM-1 polymorphism c.1405A>G (rs5498) was analyzed using PCR-restriction fragment length polymorphism, and analysis of c.721G>A (rs1799969) was done using the amplification-refractory mutation system. Further, approximately 10% of samples were validated for both polymorphisms for the observed genotypes by Sanger sequencing. A meta-analysis incorporating nine studies (1844 DR cases and 1595 controls) was also performed to assess the association of ICAM-1 rs5498 with DR risk.

RESULTS

The allele frequency and genotype distribution of ICAM-1 c.1405A>G polymorphism in the DR and control groups were not significant (P = 0.070 and P = 0.120, respectively). The GG genotype revealed a 1.6-fold increased risk of developing retinopathy (odds ratio = 1.61, 95% confidence interval: 1.01-2.58, P = 0.044). However, the AG genotype did not show any significant association (P = 0.643) between DR cases and controls. With c.721G>A in ICAM-1 the onset and progression of retinopathy was not found to be significantly correlated. The meta-analysis revealed no significant association between rs5498 and DR risk in the overall population or in Asians, but a significant association was observed in Caucasians under the allelic and recessive models.

CONCLUSION

The ICAM-1 rs5498 GG genotype increased retinopathy risk 1.61-fold in northern Indians. Meta-analysis of nine studies found no Asian association; a Caucasian signal warrants caution given limited subgroups and heterogeneity.

Key Words: Diabetic retinopathy; DNA sequencing; Intercellular adhesion molecule-1; Type 2 diabetes mellitus; Northern India

Core Tip: This study explored the association of intercellular adhesion molecule-1 gene polymorphisms (rs5498 and rs1799969) with diabetic retinopathy (DR) in patients with type 2 diabetes mellitus from northern India. A significant association was found between the rs5498 GG genotype and increased DR risk. A meta-analysis of nine studies revealed no consistent association overall or in Asians but a significant association in Caucasians. This was the first report from northern India on intercellular adhesion molecule-1 polymorphisms and DR, highlighting the importance of population-specific genetic risk assessment.

INTRODUCTION

Diabetes is the root cause of a wide range of problems, including macrovascular as well as microvascular complications. Macrovascular complications include cardiovascular, peripheral vascular, and cerebrovascular disease. Microvascular complications include neuropathy, retinopathy, and renal abnormalities. Diabetic retinopathy (DR), the most prevalent microvascular condition associated with diabetes, is the leading cause of vision impairment and blindness in persons of the working age group[1]. DR, a chronic and progressive complication, is classified into three types based on apparent ocular changes and retinal neovascularization. These types mainly include non-proliferative DR (exhibiting lesions such as microaneurysms and lipid deposits in the retina), pre-proliferative DR (primarily indicating intraretinal microvascular anomalies), and proliferative DR (where neovascularization is the distinguishing trait)[2]. DR involves a range of pathological factors, including inflammation, retinal ischemia-reperfusion injury, aberrant permeability of blood vessels, neovascularization, and macular edema. Moreover, the severity of the condition worsens with the advancement of diabetes[3].

The prevalence of retinopathy ranges from 22% to 37% among those with previously known diabetes[4]. The global prevalence of DR has been estimated to be 35%[5], and it varies among different ethnic groups[6]. In the northern Indian population, its prevalence is reported to be as high as 34%[7]. Achieving optimal glycemic control is the primary strategy in managing sight-threatening retinopathy. Reports from the Diabetes Control and Complications Trial[8] and the United Kingdom Prospective Diabetes Study[9] indicate that intensive insulin therapy and effective blood pressure regulation substantially reduce the progression of retinopathy. Strict glucose control can also delay the onset and progression of retinopathy in patients with early diabetes. Strict glucose control can delay the progression of retinopathy in early diabetes patients[10].

Studies have shown an association of systemic biomarkers like C-reactive protein, homocysteine, and advanced glycation end products with DR[11]. Elevated homocysteine/hyperhomocysteinemia levels have been reported as a risk and predictive factor for DR[12]. Homocysteine may exacerbate oxidative stress in vivo and enhance reactive oxygen species generation in the human retina[13]. Although advances in imaging, such as hyperspectral imaging, have improved early DR detection[14], such modalities do not address underlying disease susceptibility. Genetic biomarkers may therefore provide predictive value by identifying individuals at high risk before clinical manifestations appear. However, despite extensive research work in other ethnic groups, genetic studies in South Asian populations remain limited, and no published study has evaluated intercellular adhesion molecule-1 (ICAM-1) polymorphisms in patients with DR from northern India. This gap underscores the need for population-specific investigations that could complement imaging-based diagnostics and inform precision risk stratification.

Sib-pair linkage analysis and candidate gene association studies have revealed polymorphic alterations in several genes that play critical roles in various biological pathways. These include aldo-keto reductase family 1 member B1 (also known as AKR1B1), which is the first enzyme in the polyol pathway, angiotensin I-converting enzyme (also known as ACE), a key enzyme in the renin-angiotensin system, and receptor for advanced glycation end products (also known as RAGE), all of which contribute to metabolic and inflammatory processes. Additionally, complement factor H and complement factor B are involved in the inflammatory response. The gene homeostatic iron regulator is implicated in iron metabolism while superoxide dismutase 2 and pigment epithelium-derived factor are involved in oxidative stress regulation. Lastly, vascular endothelial growth factor, a growth factor, has been associated with retinopathy. Variations in these genes affect gene expression and influence how individuals respond to environmental factors, thereby playing a significant role in the pathogenesis of the disease[15].

Emerging data have emphasized the potential significance of ICAM-1 in the development of DR. ICAM-1 is a transmembrane glycoprotein that is immunoglobulin-like and expressed on the surface of leukocytes, endothelial cells, and epithelial cells[16]. It promotes immune cell migration and perivascular infiltration by influencing the adherence of circulating immune cells to the endothelium. Elevated levels of ICAM-1 and its ligands have been seen in patients with DR and retinas of animal models[17-20]. Monoclonal antibodies that block ICAM-1 efficiently prevent diabetic retinal leukostasis and vascular leakage by 48.5% and 85.6%, respectively[17]. Similarly, when the bioactivity of the ICAM-1 counter receptor CD18 is blocked, diabetic retinal leukocyte adherence is reported to be significantly reduced[18].

The ICAM-1 genetic variant c.1405A>G; p.Lys469Glu (rs5498) has been widely studied for its association with degenerative and inflammatory disorders, including DR. Studies have been conducted across various ethnic populations, including Chinese, Egyptian, Japanese, East Indian, and South Indian groups, whereas only one research report from Slovenia, focusing on the Caucasian population, has explored the association between c.721G>A; p.Gly241Arg (rs1799969) and DR on a global scale. Additionally, there is a lack of studies from South Asian populations studying the association of c.721G>A; p.Gly241Arg (rs1799969) with DR. The c.1405A>G polymorphism (rs5498) in exon 6 of ICAM-1 results in the substitution of glutamic acid for lysine at codon 469 in the immunoglobulin-like domain 5. This domain is critical as it regulates ICAM-1 binding to endothelial cells, leukocyte adhesion protein-1, and macrophage-1 antigen and hence mediates leukocytosis[21]. Since there is no published report for the ICAM-1 single nucleotide polymorphisms (SNPs) analysis with DR from the northern part of India, the present study aimed to investigate the association of rs5498 and rs1799969 with DR in the northern Indian population.

MATERIALS AND METHODS

Subjects

The study was approved by the Guru Nanak Dev University’s Ethics Committee (Date: 29 July 2013; No. 1447A/HG) in compliance with the 1975 Helsinki Declaration (2013 revision). The power of the study was calculated using the CaTS power calculator to achieve a minimum power of 80%. The effective sample size was assessed to be 270 DR cases and 270 controls. Assuming cases of nonresponse, recording mistakes, ill health, or refusal of blood sampling, the total sample size was raised by 10% for both DR cases and controls. As a result the total sample size was 302 DR cases and 312 controls aiming to investigate the association of ICAM-1 SNPs with DR.

A total of 614 participants with confirmed type 2 diabetes mellitus (T2DM) and a diabetes duration of ≥ 5 years were recruited between August 2010 and December 2014. All the participants received an ocular check-up at the Dr. Daljit Singh Eye Hospital, Amritsar, Punjab, India. Complete family histories of all the participants were collected, and each participant underwent a comprehensive ophthalmic assessment, which included visual acuity testing, Humphrey’s perimetry, ocular coherence tomography, fundus examination, and fundus photography.

Retina examinations were performed by retina specialists at the Dr. Daljit Singh Eye Hospital using binocular ophthalmoscopy following pupil dilation (with tropicamide and phenylephrine 2.5%). Fundus images were captured with a 50° angle centered on the fovea. The severity of retinopathy observed in each patient’s fundus photographs was evaluated based on the criteria established by the Early Treatment Diabetic Retinopathy Study research group[22]. Three hundred and two individuals showed signs of retinopathy and thus were classified as DR cases; 312 age-matched individuals with T2DM and no symptoms of retinopathy served as controls [controls without DR (CDR) group].

The patient group comprised individuals with T2DM and retinopathy as the major condition, a diabetes diagnosis, an age of more than 30 years, and a glycated hemoglobin (HbA1c) level ≥ 6.5%. Individuals were included in the CDR group if their random plasma glucose levels were ≥ 200 mg/dL, duration of T2DM ≥ 5 years, age at diabetes diagnosis ≥ 30 years, and if they had no microvascular or macrovascular complications. Individuals with type 1 diabetes and those with any other eye ailment, such as optic neuropathy, posterior vitreous detachment, high myopia, dense cataract, or glaucoma, were excluded from the present study. HbA1c levels were retrieved from patient records. The biochemical parameter tests [total cholesterol, triglycerides (TG), and high-density lipoprotein cholesterol (HDL-C)] were conducted following the manufacturer’s guidelines (Erba-Mannheim, TransAsia Bio-medicals Ltd., Solan, India). The calculation of low-density lipoprotein cholesterol (LDL-C) and very low-density lipoprotein cholesterol (VLDL-C) levels was performed using the formulas described by Friedewald et al[23].

Genotyping for ICAM-1 (rs5498 and rs1799969) polymorphisms

After informed consent 10 mL of peripheral venous blood was obtained from each participant, and genomic DNA was extracted using a standard phenol-chloroform method. For ICAM-1 rs5498 and rs1799969 polymorphisms, primers were designed using Primer3plus software (https://www.primer3plus.com/) and web-based allele-specific PCR software, respectively (Supplementary Table 1). A PCR followed by a restriction fragment length polymorphism technique was used to genotype rs5498 of ICAM-1. To genotype rs1799969 of ICAM-1, an amplification-refractory mutation system was performed. Further, approximately 10% of DR and CDR samples for rs5498 and rs1799969 were validated for observed genotypes using Sanger sequencing. The Exo-SAP method was used to purify the PCR products that were sequenced using the Sanger sequencing technique, and the sequencing reactions were purified using 95% isopropanol. The purified sequencing reaction products were electrophoresed using a Genetic Analyzer (Model 3500xL Genetic Analyzer; Applied Biosystems, Foster City, CA, United States). The sequences were assembled using SeqA6 sequencing analysis tools, and genotyping was done with SeqScape3 software.

Literature search and inclusion/exclusion criteria for meta-analysis

A systematic search of PubMed, EMBASE, Scopus, and Web of Science databases (January 1990-May 2025) was conducted using the keywords: Intercellular adhesion molecule-1; ICAM-1; K469E; rs5498; c.1405A>G; DR; type 2 diabetes; and polymorphism. Reference lists of selected articles and reviews were manually screened for additional studies. Only full-text, peer-reviewed publications were included with the largest cohort chosen when duplicate reports existed. Studies were eligible if they: (1) Reported ICAM-1 rs5498 genotype frequencies in DR cases and controls; (2) Were published in English or Chinese before May 2025; and (3) Provided sufficient data for calculating odds ratios (ORs) and confidence intervals (Cis). Exclusion criteria included: Case-only designs; phenotypic studies without genotyping; reviews; meta-analyses; and unpublished or gray literature.

Statistical analysis

Statistical analysis was conducted using the statistical package for the social sciences for Windows version 29 (SPSS, Inc., Chicago, IL, United States). The clinical characteristics of the DR and CDR groups were compared using unpaired Student’s t-test and one-way analysis of variance with data expressed as mean ± SD. Genotype and allele frequency distributions in the DR and CDR groups were compared using the χ² test. Binary logistic regression analyses were performed to determine the OR for the association of the rs5498 and rs1799969 polymorphisms in ICAM-1 in the DR group compared with the CDR group, adjusting for potential confounding effects across various inheritance models (dominant, codominant, and recessive). The Hardy-Weinberg equilibrium (HWE) was assessed using the Cochran-Armitage trend test (2 × 3 contingency table) through a linear regression model with P < 0.05 regarded as statistically significant. The total sample size of 614 individuals (302 DR cases and 312 controls) had a statistical power of > 90% to detect an association with an OR of 1.5 at P < 0.05. Statistical analysis for the meta-analysis was conducted using the web-based tool MetaGenyo[24]. Between-study heterogeneity was assessed using the Q statistic and I² index. Publication bias was evaluated through Egger’s test and Begg’s funnel plot.

RESULTS

A study of ICAM-1 rs5498 and rs1799969 polymorphisms with DR was conducted. The DR and CDR groups were age-matched (55.46 ± 8.25 years for DR and 55.13 ± 9.72 years for CDR). However, the mean age of onset of diabetes was significantly greater in the CDR group (P < 0.001), and the duration of diabetes was longer in the DR group (P < 0.001; Figure 1). The CDR group had higher mean systolic blood pressure values; however, the DR group had significantly higher random blood glucose levels, and the differences were statistically significant (P = 0.008 and P = 0.015, respectively). Body mass index and HbA1c level differences were also statistically significant (P = 0.012 and P = 0.017, respectively). The CDR group had significantly higher mean cholesterol (138.82 ± 58.48 mg/dL) and TG levels (201.66 ± 77.45 mg/dL), and the differences were statistically significant (P = 0.041 and P < 0.001, respectively) (Figure 1). LDL-C and VLDL-C levels were likewise higher in the CDR group, and the differences in mean values were statistically significant (P < 0.001 and P < 0.001, respectively). Statistical analysis for the association of smoking with retinopathy could not be done because all patients with DR were nonsmokers, and the number of smokers in the control group was relatively small (n = 11). Furthermore, only 22 DR cases were found to have nephropathy while no CDR participants had nephropathy; thus, no statistical analysis could be done.

Figure 1 Clinical characteristics among diabetic retinopathy cases and controls without diabetic retinopathy group (type 2 diabetes mellitus individuals without retinopathy as controls). BMI: Body mass index; A1C: Acetylated hemoglobin; HDL: High-density lipoprotein; LDL: Low-density lipoprotein; VLDL: Very low-density lipoprotein; DR: Diabetic retinopathy; CDR: Controls without diabetic retinopathy.

The allele and genotype frequency distribution of the ICAM-1 c.1405A>G (p.Lys469Glu) polymorphism among the DR and CDR groups were found to be non-significant (P = 0.070 and P = 0.120, respectively). The GG genotype was observed to be more common in the DR group compared with the CDR group (20.19% vs 14.10%) (Table 1). For the c.1405A>G polymorphism, comparing the GG genotype with the AA genotype between DR cases and controls revealed a 1.6-fold increased risk of developing retinopathy (OR = 1.61, 95%CI: 1.01-2.58, P = 0.044). However, the comparison of the AG genotype with the AA genotype between DR cases and controls revealed no significant association (OR = 1.08, 95%CI: 0.76-1.54, P = 0.643). No significant association of the c.1405A>G (p.Lys469Glu) polymorphism with retinopathy was observed under any of the analyzed genetic models (dominant, codominant, and recessive) using linear regression (Table 1). Also, DR cases were compared based on the severity of the retinopathy, and cases with proliferative retinopathy were shown to have higher frequencies of the A allele and the AA genotype for c.1405A>G polymorphism than cases with non-proliferative DR; however, the differences were not statistically significant (P = 0.585 and P = 0.627, respectively). The comparison of AG and GG genotypes with the AA genotype showed no significant association between the c.1405A>G (p.Lys469Glu) polymorphism in ICAM-1 and the progression of retinopathy (Table 2).

Table 1 Allele frequencies and genotype distribution of c.1405A>G (p.Lys469Glu) (rs5498) polymorphism in intercellular adhesion molecule-1 in diabetic retinopathy cases and controls.

	DR cases, n = 302	CDR, n = 312	χ2, P value	OR (95%CI)	OR, P value
Alleles
A	344 (56.95)	388 (62.17)	0.070	Reference
G	260 (43.04)	236 (37.82)		1.24 (0.98-1.56)	0.062
Genotypes
AA	103 (34.10)	120 (38.46)	0.120	Reference
AG	138 (45.69)	148 (47.43)		1.08 (0.76-1.54)	0.643
GG	61 (20.19)	44 (14.10)		1.61 (1.01-2.58)	0.044a
Genetic models
Dominant model (AA vs AG + GG)				0.98 (0.67-1.42)	0.916
Codominant model (AG vs AA + GG)				0.83 (0.58-1.20)	0.342
Recessive model (GG vs AG + AA)				1.31 (0.81-2.13)	0.262

^aP < 0.05.

Data are presented as n (%). Data indicate the number of subjects with each genotype and allele (frequency in percent). Odds ratios for different modes of inheritance were calculated. Odds ratios were adjusted for age, age of onset of diabetes, duration of diabetes, systolic blood pressure, diastolic blood pressure, random blood glucose levels, glycated hemoglobin, and body mass index. DR: Diabetic retinopathy; CDR: Controls without diabetic retinopathy; OR: Odds ratio; CI: Confidence interval.

Table 2 Allele frequencies and genotype distribution of c.1405A>G (p.Lys469Glu) (rs5498) polymorphism in intercellular adhesion molecule-1 in proliferative diabetic retinopathy and non-proliferative diabetic retinopathy groups.

	PDR cases, n = 189	NPDR, n = 113	χ2, P value	OR (95%CI)	OR, P value
Alleles
A	219 (57.93)	125 (55.30)	0.585	Reference	Reference
G	159 (42.06)	101 (44.69)		0.89 (0.64-1.25)	0.528
Genotypes
AA	65 (34.39)	38 (33.62)	0.627	Reference	Reference
AG	89 (47.08)	49 (43.36)		1.06 (0.62-1.80)	0.824
GG	35 (18.51)	26 (23.00)		0.78 (0.41-1.50)	0.467
Genetic models
Dominant model (AA vs AG + GG)				0.93 (0.56-1.54)	0.791
Codominant model (AG vs AA + GG)				1.15 (0.72-1.86)	0.544
Recessive model (GG vs AG + AA)				0.72 (0.40-1.30)	0.285

Data are presented as n (%). Data indicate the number of subjects with each genotype and allele (frequency in percent). Odds ratios for different modes of inheritance were calculated. Odds ratios were adjusted for age, age of onset of diabetes, duration of diabetes, systolic blood pressure, diastolic blood pressure, random blood glucose levels, glycated hemoglobin, and body mass index. PDR: Proliferative diabetic retinopathy; NPDR: Non-proliferative diabetic retinopathy; OR: Odds ratio; CI: Confidence interval.

Table 3 shows the allele frequency and genotype distribution of the ICAM-1 c.721G>A (p.Gly241Arg) polymorphism among DR cases and controls. The frequency of minor allele A was found to be higher in DR cases (3.14%) compared with the CDR group (0.32%), and these differences were statistically significant (P = 0.003). The frequency of the heterozygous genotype (GA) was observed to be higher in the DR group as compared with the CDR group (6.29% vs 0.64%). However, none of the participants exhibited the homozygous mutant genotype (AA) among both cases and controls. Consequently, no additional statistical analysis was conducted for the c.721G>A polymorphism in ICAM-1 in the analyzed DR and CDR cases. Comparing quantitative characteristics with genotypes of the c.1405A>G (p.Lys469Glu) polymorphism in ICAM-1, DR cases with the AA genotype showed a higher mean age and diabetes onset. DR cases with the GG genotype had elevated systolic blood pressure, cholesterol, HDL-C, and LDL-C levels while AG genotype cases had higher random blood glucose and TG levels. However, no statistically significant differences were observed among the genotypes for any of these analyzed variables (Table 4).

Table 3 Allele frequencies and genotype distribution of c.721G>A (p.Gly241Arg) (rs1799969) polymorphism in intercellular adhesion molecule-1 in diabetic retinopathy cases and controls.

	DR cases, n = 302	CDR, n = 312	χ2, P value	OR (95%CI)	OR, P value
Alleles
G	582 (96.35)	622 (99.67)	0.003a	Reference	Reference
A	19 (3.14)	2 (0.32)		-	-
Genotypes
GG	283 (93.7)	310 (99.35)	-	Reference	Reference
GA	19 (6.29)	2 (0.64)		-	-
AA	-	-		-	-
Genetic models
Dominant model (GG vs GA/AA)				-	-
Codominant model (GG vs GA = AA vs GA)				-	-
Recessive model (AA vs GA/GG)				-	-

^aP < 0.05.

Data are presented as n (%). Data indicate the number of subjects with each genotype and allele (frequency in percent). Odds ratios for different modes of inheritance were calculated. Odds ratios were adjusted for age, age of onset of diabetes, duration of diabetes, systolic blood pressure, diastolic blood pressure, random blood glucose levels, glycated hemoglobin, and body mass index. DR: Diabetic retinopathy; CDR: Controls without diabetic retinopathy; OR: Odds ratio; CI: Confidence interval.

Table 4 Comparison of mean values for clinical characteristics stratified by genotypes among diabetic retinopathy cases for intercellular adhesion molecule-1 c.1405A>G (p.Lys469Glu) polymorphism.

Characteristics	AA (n = 103)	AG (n = 138)	GG (n = 61)	P value
Age (years)	56.10 ± 8.22	54.93 ± 8.42	55.55 ± 7.92	0.55
Age of onset of diabetes (years)	44.17 ± 10.09	41.63 ± 9.45	42.16 ± 8.91	0.118
Duration of diabetes (years)	11.76 ± 6.27	13.02 ± 5.99	13.63 ± 5.92	0.12
Systolic blood pressure (mmHg)	132.61 ± 13.15	130.28 ± 14.82	134.69 ± 16.58	0.129
Diastolic blood pressure (mmHg)	81.42 ± 6.12	82.12 ± 6.48	82.93 ± 6.17	0.33
BMI (kg/m2)	24.58 ± 3.71	24.05 ± 3.95	23.97 ± 3.13	0.465
Random blood glucose levels (mg/dL)	198.65 ± 68.14	201.62 ± 77.81	187.92 ± 70.01	0.472
HbA1c (%)	8.13 ± 1.12	8.15 ± 1.24	7.94 ± 1.00	0.483
Cholesterol (mg/dL)	130.42 ± 56.24	126.67 ± 52.13	134.46 ± 56.07	0.634
TG (mg/dL)	154.92 ± 70.02	161.48 ± 82.99	156.78 ± 77.16	0.799
High density lipoprotein cholesterol (mg/dL)	41.67 ± 19.07	39.08 ± 21.32	45.90 ± 18.66	0.087
Low density lipoprotein cholesterol (mg/dL)	93.43 ± 61.61	91.93 ± 63.60	96.43 ± 70.28	0.902
Very low density lipoprotein cholesterol (mg/dL)	32.58 ± 15.19	33.18 ± 16.89	32.99 ± 16.74	0.961

Data are presented as mean ± SD. P values were computed by the one-way analysis of variance and indicate the significance of comparison between patients with diabetic retinopathy. BMI: Body mass index; HbA1c: Glycated hemoglobin; TG: Triglyceride.

Meta-analysis of DR and the ICAM-1 Lys469Glu variant

A total of nine studies were included in the meta-analysis, comprising 1844 patients with T2DM and DR and 1595 patients with T2DM but without retinopathy, including eight previously published studies and the current study. HWE was assessed for all included studies in the meta-analysis. Most studies exhibited HWE in both the DR and control groups with P values greater than 0.05, indicating no significant deviation from equilibrium. However, the study by Zhou et al[29] in the Chinese population showed slight deviations (HWE P = 0.039)[25-32] (Table 5). Importantly, the present study from the northern Indian population was in strong HWE (P = 0.879), supporting the reliability of the genotyping results. These findings suggest that the genotype distributions in the included studies are generally consistent with HWE expectations, reinforcing the validity of the meta-analysis.

Table 5 Characteristics of the studies assessing the association between intercellular adhesion molecule-1 Lys469Glu variant and the risk of diabetic retinopathy.

No.	Ethnicity	Country	Genotyping	Diabetic retinopathy			Non-diabetic retinopathy			HWE P values	Ref.
				AA	AG	GG	AA	AG	GG
1	Asian	Japan	PCR-RFLP	34	35	12	10	30	10	0.157	Kamiuchi et al[30], 2002
2	Asian	China	PCR-HA SSCP	81	40	11	16	15	9	0.152	Liu et al[27], 2006
3	Caucasian	Slovenia	PCR-RFLP	47	96	52	44	77	22	0.218	Petrovic et al[31], 2008
4	Asian	India	Snapshot	103	162	80	99	174	86	0.578	Balasubbu et al[32], 2010
5	Asian	China	Allele specific PCR	10	24	6	24	14	2	0.982	Zhu et al[28], 2010
6	Asian	China	PCR-RFLP	37	44	21	54	45	21	0.039	Zhou et al[29], 2010
7	Asian	India	DNA sequencing	60	92	47	29	84	44	0.317	Vinita et al[26], 2012
8	Asian	China	PCR-LDR	221	193	34	166	154	24	0.142	Lv et al[25], 2016
9	Asian	India	PCR-RFLP	103	138	61	120	148	44	0.879	Present study

HWE: Hardy-Weinberg equilibrium; PCR-RFLP: PCR-restriction fragment length polymorphism; PCR-HA SSCP: PCR-heteroduplex-single-strand conformation polymorphism; PCR-LDR: PCR-ligase detection reaction.

Subgroup analysis based on ethnicity was conducted to evaluate potential differences in the association between the ICAM-1 Lys469Glu (K469E) polymorphism and DR among Asian and Caucasian populations. In the allelic model (G vs A) (Figure 2A), no significant association was observed in the overall population (OR = 0.99, 95%CI: 0.79-1.26, P = 0.987) or in the Asian subgroup (OR = 0.94, 95%CI: 0.74-1.21, P = 0.679). However, a significant association was found in the Caucasian population (OR = 1.44, 95%CI: 1.06-1.95, P = 0.021) (Table 6). Similarly, under the recessive model (GG vs AG + AA) (Figure 2B), no significant association was observed in the overall group (OR = 1.08, 95%CI: 0.80-1.47, P = 0.623) or in the Asians (OR = 0.99, 95%CI: 0.74-1.34, P = 0.961), but a significant association was noted in Caucasians (OR = 2.00, 95%CI: 1.15-3.48, P = 0.014) (Table 6). In the dominant model (AG + GG vs AA) (Figure 2C), no significant associations were detected in the overall population (OR = 0.951, 95%CI: 0.69-1.31, P = 0.951) or in Asians (OR = 0.90, 95%CI: 0.64-1.28, P = 0.566). Although the effect estimates in Caucasians suggested a possible association (OR = 1.40, 95%CI: 0.86-2.27), it did not reach statistical significance (P = 0.173) (Table 6). These possible associations observed in Caucasians cannot be considered conclusive due to the limited number of studies and insufficient statistical power. Substantial heterogeneity was observed across studies in all genetic models (I² = 76%-88%). The funnel plot appeared symmetrical (Figure 3), and Egger’s test did not indicate any evidence of publication bias (Table 6).

Figure 2 Forest plots for included studies evaluating the association between intercellular adhesion molecule-1 c.1405A>G (rs5498) polymorphism and type 2 diabetic retinopathy. A: Allelic model (G vs A); B: Recessive model (GG vs GA + AA); C: Dominant model (GG + GA vs AA). DR: Diabetic retinopathy; CDR: Controls without diabetic retinopathy; OR: Odds ratio; CI: Confidence interval.

Figure 3 Funnel plots analysis for included studies for evaluating the publication bias of the association between intercellular adhesion molecule-1 c.1405A>G (rs5498) polymorphism and type 2 diabetic retinopathy. A: Allelic model (G vs A); B: Recessive model (GG vs GA + AA); C: Dominant model (GG + GA vs AA). DR: Diabetic retinopathy; CDR: Controls without diabetic retinopathy.

Table 6 Meta-analysis results of intercellular adhesion molecule-1 Lys469Glu polymorphism under different genetic models in overall and Asian populations.

Model		Allelic model (G vs A)			Recessive model (GG vs AG + AA)			Dominant model (AG + GG vs AA)
Ethnicity		Overall	Asian	Caucasian	Overall	Asian	Caucasian	Overall	Asian	Caucasian
Number of studies		9	8	1	9	8	1	9	8	1
Test of association	Model	Random	Random	Fixed	Random	Random	Fixed	Random	Random	Fixed
	OR (95%CI)	0.99 (0.79-1.26)	0.94 (0.74-1.21)	1.44 (1.06-1.95)	1.08 (0.80-1.47)	0.99 (0.74-1.34)	2 (1.15-3.48)	0.951 (0.69-1.31)	0.90 (0.64-1.28)	1.40 (0.86-2.27)
	P value	0.987	0.679	0.021	0.623	0.961	0.014	0.951	0.566	0.173
Test of heterogeneity	I2	0.79	0.79	NA	0.57	0.48	NA	0.76	0.77	NA
	P value	< 0.001	< 0.001	NA	0.02	0.06	NA	< 0.001	< 0.001	NA
Egger’s test P value		0.866	0.796	NA	0.852	0.759	NA	0.857	0.782	NA

OR: Odds ratio; CI: Confidence interval.

Sensitivity analysis

A sensitivity analysis was conducted by sequentially omitting one study at a time to assess the robustness of the meta-analysis findings. The results demonstrated that no single study had a disproportionate impact on the overall effect estimates under the allelic (G vs A) (Figure 4A), recessive (GG vs GA + AA) (Figure 4B), and dominant (GG + GA vs AA) (Figure 4C) genetic models. The pooled ORs and CIs remained stable across all iterations, confirming the reliability and consistency of the meta-analysis results.

Figure 4 Sensitivity analysis for diabetic retinopathy vs controls without diabetic retinopathy, eliminating one study at a time. A: Allelic model (G vs A); B: Recessive model (GG vs GA + AA); C: Dominant model (GG + GA vs AA). DR: Diabetic retinopathy; CDR: Controls without diabetic retinopathy; OR: Odds ratio; CI: Confidence interval.

DISCUSSION

The present study investigated the association between ICAM-1 gene polymorphisms (rs5498 and rs1799969) and DR in patients with T2DM from northern India. Our study revealed statistically significant variances in the mean age of diabetes onset, duration of diabetes, systolic blood pressure, random blood glucose, and HbA1c levels (P < 0.05). Additionally, we investigated the correlation between lipid profile parameters (cholesterol, TG, HDL-C, LDL-C, and VLDL-C) and DR. We found statistically significant differences (P < 0.05) in body mass index, total cholesterol, TG, LDL-C, and VLDL-C values between DR and CDR cases. Hedge and Vekategowda[33] also found a strong association between high cholesterol and retinopathy in the southern Indian population. Similarly, Klein et al[34] revealed a statistically significant association between cholesterol levels and proliferative DR in their investigation on a population from southern Wisconsin in the United States. However, HDL-C levels are protective against retinopathy[34]. In contrast our study did not observe a significant association between HDL-C levels and DR. Additionally, a lack of association between cholesterol and TG levels with retinopathy was reported in a Chinese population with T2DM[15,35].

In this study a meta-analysis was performed using data from nine studies involving Asian and Caucasian populations. The analysis found no significant association between the ICAM-1 K469E (rs5498) polymorphism and DR in the overall population. This result was consistent across all genetic models including allelic, recessive, and dominant. Subgroup analyses by ethnicity showed a significant association in Caucasian populations under the allelic (G vs A) and recessive (GG vs AG + AA) models while no such association was observed in Asian populations, including our analyzed cohort in the present study. However, these findings in Caucasians were based on limited studies, and the statistical power was insufficient to draw definitive conclusions. The small number of studies in certain subgroups (e.g., Caucasians) and lack of uniformity in study designs may limit the robustness of meta-analysis findings. The funnel plot analysis indicated no substantial publication bias, and Egger’s test supported this observation, confirming the reliability of the meta-analysis results. Sensitivity analysis further demonstrated that no single study disproportionately influenced the overall effect estimates, reinforcing the robustness of the meta-analysis.

Interestingly, the present study identified a significant association between the ICAM-1 rs5498 GG genotype and DR risk in the northern Indian population despite the meta-analysis not supporting a consistent association across Asian populations. This discrepancy may be attributed to several factors. First, our study focused on a specific ethnic group (Northern Indians), who may harbor unique genetic backgrounds, environmental exposures, or gene-gene interactions not captured in other Asian cohorts. Additionally, sample size, selection criteria, and phenotypic heterogeneity across studies could contribute to the lack of a consistent association in the meta-analysis. The potential population-specific effects of the ICAM-1 polymorphism highlight the need for larger, well-powered, and ethnically diverse studies to validate these findings. Taken together, our findings highlight the need for genetic biomarkers that can aid in the early identification of individuals at higher risk of DR. In regions with high disease prevalence, such biomarkers may enhance clinical decision-making along with complement emerging technologies like hyperspectral imaging, which allow noninvasive detection of retinal changes[14]. Integrating genetic and imaging biomarkers could ultimately improve risk prediction and monitoring strategies for DR.

The association of two SNPs, c.1405A>G (p.Lys469Glu) and c.721G>A (p.Gly241Arg), in the coding region of ICAM-1 has been investigated in relation to various microangiopathies. These include DR and immune-mediated diseases such as Graves’ disease, Behcet’s disease, inflammatory bowel disease, coronary artery disease, and type 1 diabetes across different population groups worldwide[30,36,37]. The present study on the northern Indian population observed a statistically significant association of the GG genotype of the c.1405A>G (p.Lys469Glu) polymorphism in ICAM-1 with the development of DR (OR = 1.61, 95%CI: 1.01-2.58, P = 0.044). However, no significant association was observed between the c.721G>A (p.Gly241Arg) polymorphism and the development or progression of the retinopathy. The lack of association with the c.721G>A may be due to the very small number of patients and controls with the heterozygous (GA) genotype, and the absence of the homozygous mutant (AA) genotype in any of the analyzed cases and controls. Similar findings have been reported in the Caucasian population with T2DM in which no significant association of the c.721G>A (p.Gly241Arg) polymorphism with retinopathy was observed[31].

According to Vinita et al[26] the A allele of the c.1405A>G (p.Lys469Glu) polymorphism was associated with a 1.8 times increased risk of DR in the south Indian population. Similarly, Liu et al[27] and Kamiuchi et al[30] reported that the AA genotype of the ICAM-1 polymorphism (c.1405A>G) (p.Lys469Glu) was associated with a two-fold increased risk of developing retinopathy in Chinese and Japanese populations with T2DM, respectively. However, Balasubbu et al[32] reported no significant association between p.Lys469Glu and retinopathy in a study on the south Indian population. Additionally, Lv et al[25] and Yang et al[38] observed no association between p.Lys469Glu and DR in Chinese cohorts with T2DM, demonstrating a difference from the findings in the present study. Furthermore, Fan and Liu[39] reported a lack of association between the ICAM-1 polymorphism c.1405A>G and type 2 DR in the Asian population. In a meta-analysis, Sun et al[40] documented no evidence of an association between the c.1405A>G (p.Lys469Glu) polymorphism with DR. Similarly, Xie and Liang[41] conducted a meta-analysis and reported no association between the c.1405A>G polymorphism in ICAM-1 and DR.

The precise role of the polymorphisms (c.1405A>G and c.721G>A) in ICAM-1 in diabetes and its related complications remains unclear. However, it is hypothesized that amino acid substitutions in ICAM-1 could influence the three-dimensional conformation of the protein and its serum concentration. Yao et al[42] performed a meta-analysis and reported that increased levels of ICAM-1 are generally present in patients with DR and may be associated with the severity of DR. ICAM-1 proteins serve as ligands for integrins, which act as primary receptors facilitating cell-cell interactions and signal transduction. Unlike many integrin-binding proteins, ICAM-1 lacks the Arg-Gly-Asp (RGD) motif that promotes integrin binding[43]. However, ICAM-1 binds to two integrins of the β2 subunit family, namely leukocyte adhesion protein-1 and macrophage-1 antigen[44]. The aggregation of leukocytes and their subsequent adherence to the endothelium leads to injury in endothelial cells. Through its interaction with integrins, ICAM-1 plays a crucial role in immune-related functions, including the activation of T lymphocytes and the mediation of leukocyte-endothelial cell interactions[45,46]. ICAM-1 plays a significant role in the anti-inflammatory pathway through intricate interactions. Experimental studies have shown that inhibiting ICAM-1 expression can lead to an improvement in the progression of diabetes and its related complications[16]. In diabetic rats the glucagon-like peptide 1 receptor agonist exendin-4 has been shown to reduce the expression of ICAM-1[47].

In a study on the Caucasian population, Ponthieux et al[48] reported lower concentrations of ICAM-1 in healthy participants (413 children aged 6-21 years and 363 adults aged 38-55 years) with the AA genotype of the c.721G>A (p.Gly241Arg) variation as compared with those with the GG genotype. In a study on Malaysians with T2DM, Abu-Seman et al[37] reported a significant association of the p.Lys469Glu polymorphism with diabetic neuropathy. Luo et al[49] and Cui et al[50] documented a significant association between the p.Lys469Glu polymorphism and coronary heart disease in the Chinese population. Apart from diabetes-related complications, the role of ICAM-1 polymorphisms in other diseases such as urothelial cell carcinoma, childhood asthma, celiac disease, and various cancers has also been reported[51-55]. Variations in study results across different groups may stem from factors like sample size, population structure, ethnicity, genetic, and environmental differences. Genetic association studies are susceptible to statistical errors, and population-based genotypes may also have diverse effects in various populations.

Recent advances since 2020 have provided new perspectives on ICAM-1 biology relevant to DR. Single-cell RNA sequencing and multiomics studies have identified specific retinal endothelial and microglial subpopulations with markedly elevated ICAM-1 expression during early disease stages, offering unprecedented cell-type resolution[56,57]. Epigenetic investigations have revealed promoter DNA methylation changes, histone modifications, and microRNA-mediated repression (e.g., miR-146a) that govern ICAM-1 expression under hyperglycemic stress, supporting the concept of “metabolic memory”[58,59]. In addition, recent preclinical therapeutic studies have highlighted the translational potential of ICAM-1 inhibition. Notably, classic antibodyblockade studies in diabetic rats demonstrated that anti-ICAM-1 treatment suppresses retinal leukostasis and vascular leakage, and recent in vivo labeling work shows robust induction of luminal endothelial ICAM-1 in diabetic retina, together supporting the translational potential of ICAM-1 inhibition[17,60,61]. These insights enhance the contemporary relevance of our findings and underline ICAM-1 as both a biomarker and a potential therapeutic target in DR.

Strengths and limitations

The present study reported the association of the c.1405A>G (p.Lys469Glu) polymorphism in ICAM-1 with DR in a northern Indian population. Some limitations of the present study are the potential for selection bias inherent in hospital-based recruitment. Firstly, the heterogeneity of our clinically recruited population, encompassing varied genetic backgrounds and potential confounding factors, may have influenced the observed associations and limited the generalizability of our findings.

Secondly, financial constraints have restricted our ability to correlate the ICAM-1 genotype with its expression level. Future studies should incorporate serum/plasma collection and proteomic analyses, such as ELISA and western blot assays, and enable integration of genomic and proteomic datasets to better understand the functional implications of these polymorphisms. The literature also indicates that microRNAs (e.g., miR-141) regulate ICAM-1 expression, but the interactions between ICAM-1 polymorphisms (rs5498, rs1799969) and microRNAs remain unexplored; investigating these could provide important mechanistic insights[62].

Thirdly, our study did not address potential gene-environment interactions or other hereditary and environmental factors, such as diet, glycemic control, comorbidities, and additional genetic variants within immune-related genes, which may influence DR risk and progression. Future studies exploring broader genetic heterogeneity, including other immune-related gene polymorphisms, may be important to better understand DR susceptibility.

Lastly, the power of the study was calculated to be 80% with 302 DR cases and 312 controls. However, future research may involve larger, ethnically diverse cohorts, integrated genomic and functional analyses. In addition, the application of multiple testing corrections is needed to minimize false positives, enhance generalizability, and increase power to detect small effect sizes. These approaches will provide a more comprehensive understanding of the underlying mechanisms. Additionally, as this is a cross-sectional study, it restricts our ability to establish temporal or causal relationships between ICAM-1 polymorphisms and DR incidence or progression. Longitudinal cohort studies will be essential to determine whether these polymorphisms contribute causally to disease development or serve as markers of disease susceptibility.

CONCLUSION

The present study identified the c.1405A>G (p.Lys469Glu) polymorphism in ICAM-1 to be significantly associated with the development of DR in northern Indian patients compared with patients with T2DM but without DR. This genetic variation could serve as a potential genetic marker for susceptibility to retinopathy development, particularly in this population. These findings could hold clinical significance and warrant further research, including cross-ethnicity studies and functional investigations, to clarify their role in diabetic complications. Although our meta-analysis provided a comprehensive assessment of the association between ICAM-1 K469E and DR, the substantial heterogeneity across studies suggests that the pooled estimates should be interpreted with caution. Future standardized and population-specific studies are essential to validate these results.

ACKNOWLEDGEMENTS

The authors thank the study participants for their cooperation and participation.