HLA allelic variation and COVID-19 severity in a Kuwait-based population: a high-resolution immunogenetic study

Human leukocyte antigen (HLA) polymorphisms play a vital role in antiviral immunity, but their influence on COVID-19 severity differs significantly among ethnic groups. T-cell responses are essential for clearing SARS-CoV-2, yet the immunogenetic factors underlying the diverse clinical outcomes in COVID-19, especially in certain populations, are not fully understood. In a Kuwait-based cohort from national referral and public health facilities, we examined HLA allelic variants to identify potential protective or risk factors related to COVID-19 severity.

We studied 494 participants: 75 healthy controls and 419 COVID-19 patients, classified by severity (Recovered, Moderate, Severe, Critical/ICU). HLA alleles at nine loci (HLA-A, -B, -C, -DRB1, -DRB3/4/5, -DQA1, -DQB1, -DPA1, -DPB1) were genotyped via next-generation sequencing. Associations with severity were tested using logistic regression, adjusting for age, sex, and ABO blood group, with the Severe/ICU group as the reference. This was repeated, excluding controls from the reference, to separate severity effects from susceptibility to infection.

HLA genotype was successfully linked to complete clinical/covariate data for 364 of 494 participants (74%; 60 Healthy, 77 Recovered, 125 Moderate, 66 Severe, 36 ICU), though linkage success was lower among ICU patients (47%) than other strata (75-82%), a difference we consider more likely attributable to attrition during critical care than to allele-related bias. Among these 364 participants, 22 HLA alleles spanning all nine genotyped loci showed nominal association with severe/critical disease (P < 0.10) in the primary model, predominantly risk-conferring: HLA-C*06:02:01 (OR 2.50, 95% CI 1.43-4.36, P = 0.001), HLA-DQB1*02:02:01 (OR 2.25, 1.31-3.88, P = 0.003), HLA-DRB1*15:03:01 (OR 4.65, 1.38-15.67, P = 0.013), HLA-A*29:02:01 (OR 14.11, 1.43-138.81, P = 0.023), and HLA-B*13:02:01 (OR 12.78, 1.31-124.18, P = 0.028), among others; one protective association was identified, HLA-A*31:01:02 (OR 0.26, 0.08-0.83, P = 0.023). These five associations remained directionally concordant in the sensitivity analysis excluding Healthy controls, whereas several weaker associations did not. None survived correction for multiple testing (FDR-adjusted q ranged from 0.36 to 0.99); a minimum-detectable-odds-ratio calculation indicated this cohort was powered to detect only large effects (OR ≈ 10 for a 1%-frequency allele after correction), consistent with underpowering rather than absence of true effect for several low-frequency candidates. Age remained the strongest independent predictor of severity (P < 0.001), whereas sex showed no significant association after adjustment (P = 0.289).

This study provides a comprehensive characterisation of HLA allele frequencies in a Kuwait-based population, including their association with COVID-19 clinical outcomes. Age was a substantially more dominant determinant of COVID-19 severity than HLA genotype in this cohort. None of the identified HLA alleles retained statistical significance after correction for multiple testing; given both this power limitation and the absence of individual-level ancestry and vaccination data in a population with substantial genetic and demographic heterogeneity, these findings should be regarded as hypothesis-generating candidates for replication in larger, ancestry-adjusted, adequately powered cohorts.