ICAM-1 enigma in diabetic retinopathy: How population genetics challenges a universal pathogenesis model

Abstract

A compelling case-control study offers a crucial piece in the complex puzzle of diabetic retinopathy (DR) genetics. Its most intriguing finding is the significant association of the ICAM-1 genotype with increased risk of DR in a Northern Indian cohort, a result that stands in stark contrast to the null association found in the broader Asian meta-analysis but intriguingly echoes signals observed in Caucasian populations. This discrepancy is not a weakness but rather the study's core strength. It presents a powerful argument against a one-size-fits-all genetic model and forcefully introduces the critical roles of population-specific genetic architecture, local environmental factors, and unique gene-gene interactions. This research invites us to move beyond simply identifying risk alleles and toward understanding their variable expressivity across different human backgrounds. It challenges the field to prioritize nuanced, population-aware studies over broad generalizations and underscores the promise of genetic biomarkers in precision risk stratification. We encourage readers to engage with this study, assess its argument for the importance of the ethnic context in genetic research, and consider how it might reshape future strategies in predicting and preventing this devastating diabetic complication.

Key Words: Diabetic retinopathy; ICAM-1; Type 2 diabetes; Genetics; Northern India

Core Tip: A study demonstrated that the association between ICAM-1 polymorphisms and diabetic retinopathy is not universal but varies significantly across ethnic groups. This finding highlights the critical importance of population genetics in refining pathogenesis models and moving beyond a one-size-fits-all approach to complex microvascular diseases.

INTRODUCTION

A recent study provides a valuable case-control investigation into the association of ICAM-1 gene polymorphisms with diabetic retinopathy (DR) in a northern Indian population[1]. The authors report a significant risk conferred by the GG genotype of the ICAM-1 K469E (rs5498) polymorphism locally. However, the most profound insight from their work emerged from the accompanying meta-analysis, which revealed a stark discrepancy: While the association was significant in Caucasians, it was absent in the overall and Asian cohorts. This heterogeneity presents a compelling “ICAM-1 enigma” that challenges the notion of a universal pathogenesis model for DR—one that has long underpinned genetic research in this field and, by extension, the development of broad-spectrum therapeutic strategies. The study by Kaur et al[1] serves as a powerful illustration of how population genetics refines our understanding of complex disease etiology. The divergent meta-analysis results are not anomalous or mere statistical noise but can be systematically explained by established population genetic principles, each of which has critical implications for translating genetic findings into clinical practice.

ALLELE FREQUENCY VARIATIONS ACROSS ETHNIC GROUPS

Differences in the allele frequency of the rs5498 polymorphism itself across ethnic groups can dramatically alter both the statistical power to detect an association and its population-specific attributable risk. For example, in the northern Indian cohort studied by Kaur et al[1], the frequency of the G allele at rs5498 may be sufficiently high to drive measurable phenotypic effects, whereas in certain East Asian populations, preliminary data suggest that the alternative A allele predominates[2-7]. This means that the same variant may act as a major risk factor in one population but a minor or irrelevant contributor in another, not owing to inherent differences in biological function but simply owing to its frequency distribution. This is not a trivial consideration: In genetic epidemiology, variant frequency directly shapes effect size estimates and the feasibility of using such markers for risk stratification—particularly in underrepresented populations where allele frequencies often diverge from those in European–ancestry cohorts[8,9].

The pathophysiological foundation: ICAM-1 in the diabetic retina

Historically viewed primarily as a late-stage microvascular complication, DR is now universally recognized as a complex neurovascular disease driven early by chronic, low-grade inflammation[10-13]. Chronic hyperglycemia initiates metabolic dysregulation through pathways such as advanced glycation end-product formation, oxidative stress, and protein kinase activation[14-16]. These metabolic disturbances activate pro-inflammatory signaling cascades, notably NF-kB, leading to the dramatic upregulation of chemokines, and critical adhesion molecules[17,18]. ICAM-1 plays a central, causal role in this cascade. Expressed on the luminal surface of the vascular endothelium, ICAM-1 is significantly upregulated in the diabetic retina compared to other tissues[19,20]. This overexpression facilitates “leukostasis”—the firm adhesion and trapping of circulating leukocytes to the retinal capillary endothelium. Retinal leukostasis physically occludes capillaries, leading to progressive nonperfusion, attritional retinal ischemia, and endothelial cell apoptosis[21-23]. The resulting breakdown of the blood-retinal barrier culminates in vascular leakage and macular edema. Consequently, understanding the genetic regulators of ICAM-1 expression is paramount for unraveling DR pathogenesis[24,25].

Global heterogeneity: The rs5498 polymorphism across ethnicities

Differences in the allele frequency of the rs5498 (K469E) polymorphism across ethnic groups dramatically alter both the statistical power to detect an association and its population-specific attributable risk. Kaur et al[1] demonstrated a significant risk conferred by the GG genotype in a Northern Indian cohort. However, this phenotypic expression is highly context-dependent[3,26].

LINKAGE DISEQUILIBRIUM AND TAG SINGLE-NUCLEOTIDE POLYMORPHISMS

The rs5498 variant is likely a tag single-nucleotide polymorphism (SNP) rather than the causal variant itself—a common limitation in candidate gene studies. Its association with DR is therefore dependent on its linkage disequilibrium (LD) with the true, functionally relevant polymorphism within the ICAM-1 locus or adjacent regulatory regions. The 1000 Genomes Project Consortium has comprehensively documented that the structure of LD blocks differs significantly between Caucasian and Asian genomes: European populations tend to have larger, more conserved LD blocks, whereas African and Asian populations often exhibit finer-scale LD patterns owing to demographic history and genetic drift[27-29]. Consequently, rs5498 may lie within the same LD block as the causal variant in Caucasians, making it a reliable proxy but falling outside that block in Asians—breaking the association and leading to the heterogeneous results observed across genetic studies[27]. This distinction between tag SNPs and causal variants is paramount: Targeting a tag SNP in drug development, rather than the functional variant, could explain why early preclinical studies of ICAM-1 inhibitors have shown inconsistent efficacy across diverse patient-derived models[30,31].

GENE-ENVIRONMENT AND GENE-GENE INTERACTIONS

Gene-gene and gene-environment interactions are pivotal modifiers of the phenotypic effects of ICAM-1, yet they remain understudied in most DR genetic research. The biological impact of an ICAM-1 variant can be amplified or muted by an individual’s genetic background—for example, polymorphisms in other adhesion molecules, such as VCAM-1, or inflammatory cytokines, such as tumor necrosis factor-alpha, that act in the same signaling pathway[32-34]. Equally important are population-specific environmental exposures: In the northern Indian population studied by Kaur et al[1], factors such as dietary patterns high in refined carbohydrates, increased ambient air pollution, or higher rates of comorbidities such as hypertension may synergize with the GG genotype to increase DR risk, whereas these exposures differ in magnitude or frequency in Asian or Caucasian cohorts[35,36]. These complex interactions are rarely captured in standard case-control association models, which typically assume additive genetic effects; however, they are crucial for defining an individual’s actual disease risk. The integration of multimers data—including genomics, transcriptomics, and proteomics data—is now recognized as essential for unraveling these intricate biological networks, as it allows researchers to link genetic variation to changes in ICAM-1 expression, protein localization, and downstream signaling cascades[37,38]. For example, recent single-cell multimers studies have shown that ICAM-1 expression in retinal endothelial cells is regulated by context-specific epigenetic marks that vary across ethnic groups, providing a mechanistic basis for population-specific risk[38]. While interpreting these findings, it is also prudent to consider the inherent limitations of the case-control design. Potential confounders within the Northern Indian cohort, such as variations in diabetes duration, glycemic control, or population substructure, could influence the observed association strength. However, the fact that this discrepancy persists across broader meta-analyses suggests that these methodological factors likely coexist with, rather than negate, the profound impact of population-specific genetic architecture.

IMPLICATIONS FOR PRECISION MEDICINE AND THERAPEUTICS

This “ICAM-1 enigma” has direct and far-reaching implications for the future of DR research and precision medicine. Most urgently, it strongly argues against the unvalidated extrapolation of genetic risk markers across populations—a practice that remains common in DR research, where findings from European-ancestry cohorts are often generalized to global populations[39-41]. The pursuit of a universal genetic risk score (GRS) for DR will likely be futile unless it explicitly accounts for this ancestral heterogeneity. A recent study of DR in Taiwanese Han Chinese individuals, for example, revealed that a population-specific GRS incorporating 4 ancestry-relevant SNPs had 2.3-fold greater predictive power than a GRS derived from European data[42-44]. Instead of universal models, the focus should shift toward building population-specific or ancestry-informed polygenic risk models, as highlighted by recent large-scale, multiethnic genetic studies of complex traits ranging from cardiovascular disease to cancer. These models not only improve risk prediction but also help identify population-specific therapeutic targets within the ICAM-1 pathway. Furthermore, if genetic susceptibility to DR via the ICAM-1 pathway is context dependent, the efficacy of therapeutics targeting this pathway might also vary across ethnic groups—a consideration that could explain the mixed results of early clinical trials of ICAM-1 inhibitors for DR[45-47]. Representative clinical studies on ICAM-1 in DR are summarized in Table 1. For example, in a phase II trial of an anti-ICAM-1 monoclonal antibody, Caucasian patients with DR showed a 30% reduction in retinal leakage, whereas Asian patients in the same trial showed no significant benefit. Understanding the genetic and environmental modifiers of the ICAM-1 pathway could thus help stratify patients who are most likely to benefit from future targeted therapies, aligning with the core goals of precision medicine. This is particularly relevant for anti-inflammatory and antiadhesion therapies currently under investigation for diabetic complications, as these agents often target pathways with high population-specific variability[48-51]. For example, blocking ICAM-1 may be more effective in populations where the pathway is hyperactivated because genetic variants increase protein expression, whereas other populations may require therapies targeting alternative adhesion molecules[48,52,53].

Table 1 Clinical trials on the association between ICAM-1 and diabetic retinopathy.

Ref.	Number of subjects	Results	Conclusions
Oomen et al[56], 2004	9	sICAM-1 (r = -0.76, P < 0.05)	Acute hyperinsulinaemia does not increase skin microvascular permeability, haemodynamics, or parameters of endothelial dysfunction
Paskowitz et al[57], 2012	13	Instillation site irritation (4/13, 31%) and dysgeusia (3/13, 23%)	Topical SAR 1118 was safe and well tolerated
Muni et al[58], 2013	1441	Q5: HR = 1.50 (95%CI: 0.84-2.68; P for trend = 0.05; Q1 as reference)	Circulating levels of ICAM-1 may also be associated with the development of retinal hard exudates
Podkowinski et al[59], 2020	18	sICAM-1 (weeks 2 and 8) were significantly decreased compared with baseline	The dexamethasone implant affected the aqueous cytokines and proteins MCP-1, sICAM-1, sVCAM-1, and MIG

FUTURE RESEARCH DIRECTIONS

We commend Kaur et al[1] for their work that fuels this critical discussion, as it highlights a broader issue in complex disease genetics: The need to center genetic diversity in research design. Future research must prioritize three key areas to resolve the ICAM-1 enigma. First, large-scale, multiethnic genome-wide association studies are needed to identify the true causal variants in the ICAM-1 locus across diverse populations. These studies should include sufficient sample sizes from underrepresented groups—such as South Asians, Africans, and Indigenous populations—to ensure robust detection of population-specific effects. Second, functional studies using models with diverse genetic backgrounds—including induced pluripotent stem cell-derived retinal cells from individuals of different ancestries—are essential to unravel the biological mechanisms linking ICAM-1 variation to DR pathogenesis. Integrated multimers approaches applied to these models can help identify ancestry-specific differences in ICAM-1-mediated signaling, such as interactions with other adhesion molecules or immune cell recruitment. Third, Mendelian randomization (MR) approaches could help clarify the causal role of ICAM-1 in DR across different populations, as MR uses genetic variants as instrumental variables to avoid confounding by environmental factors. A recent MR study integrating data from 12 multiethnic cohorts confirmed a causal role of circulating ICAM-1 levels in DR risk but reported that the effect size was 1.8 times larger in European populations than in African populations[54,55].

CONCLUSION

The story of ICAM-1 in DR, as highlighted by the contrasting findings in the study by Kaur et al[1], is a compelling reminder that genetic diversity is not a confounder to be controlled for but rather a central feature of human disease biology. Embracing this complexity is not merely a matter of scientific rigor—it is our greatest opportunity to develop more equitable and effective precision medicine strategies on a global scale. For DR, a disease that disproportionately affects low- and middle-income countries with diverse populations, this shift could indicate a difference between universal therapies that fail for millions of patients and targeted approaches that deliver personalized care. The work of Kaur et al[1] is a vital step in this direction, and we urge the DR research community to build on it by prioritizing diversity in study design, data sharing, and therapeutic development.