Association of GBP2 and GBP5 expression with COVID-19 severity: a retrospective cross-sectional study

Abstract Background Host interferon responses are increasingly recognized as determinants of COVID-19 outcomes. Guanylate-binding proteins (GBPs) are interferon-stimulated GTPases that restrict SARS-CoV-2 by interfering with furin-mediated processing of the viral spike protein; however, their relationship with clinical severity in patients has not been extensively studied. This study evaluated the association between GBP2 and GBP5 mRNA expression and clinical severity, with GBP5 as a comparator to test whether any association was gene-specific. Methods In this retrospective, cross-sectional study of archived whole-blood samples collected between November 2020 and February 2021, 150 adults with RT-PCR-confirmed SARS-CoV-2 infection (37 mild, 32 moderate, 46 severe, 35 critical, classified per WHO/national criteria) and 50 SARS-CoV-2-negative controls were assayed for GBP2 and GBP5 mRNA by qRT-PCR, normalized to GAPDH (2 − ​ ΔΔCt method). Groups were compared using one-way ANOVA with Tukey post-hoc testing, with variance-robust and log2-scale sensitivity analyses; age, sex, and comorbidity were assessed by Spearman correlation and regression. Results GBP2 expression differed significantly across the five study groups (F = 288.27, p < 0.001), decreasing stepwise from mild (4.76-fold) to critical (0.60-fold) disease relative to controls. All ten pairwise comparisons were significant on the fold-change scale; on the log2 scale nine were significant with Tukey HSD (mild versus moderate, p = 0.205) and all ten with Games–Howell. GBP2 was inversely correlated with age ( r = − 0.57, p < 0.001), but clinical severity remained an independent predictor after adjustment ( p < 0.001 for each severity group). Sex and comorbidity were not independently associated, and age reached significance on the log2 scale only (conventional p = 0.045; HC3 p = 0.020). GBP5 differed far less across groups (F = 3.31, p = 0.012), with higher expression in critical patients than controls (1.35-fold vs. 1.00-fold) and no graded pattern. The crude association between low GBP2 and mortality was no longer detectable after adjusting for severity. Conclusions GBP2 expression was inversely associated with COVID-19 severity in this cohort, showing a strong graded relationship across severity strata, whereas GBP5 showed a much weaker and non-graded pattern. As only these two interferon-stimulated genes were measured, no inferences can be drawn about such genes in general. These findings describe an observed cross-sectional association, not a causal one, and warrant confirmation in prospective multicenter studies.