Association between serum IgA levels and mortality in patients with sepsis and serum IgG ≥670 mg/dL: an observational study of clinical and MIMIC-IV cohorts

Background The association between serum immunoglobulin concentrations and mortality in sepsis remains uncertain. This study examined the association between serum IgA and mortality in patients with sepsis and serum IgG ≥670 mg/dL. Methods Based on real-world clinical data from ICUs at two tertiary Class A hospitals in mainland China and the MIMIC-IV database, this study divided patients into IgG <670 mg/dL and IgG ≥670 mg/dL groups using a serum IgG concentration of 670 mg/dL as the cutoff. This study used comparative analysis, Cox regression, and Log-rank survival analysis to examine the association between serum IgA and mortality in sepsis patients with serum IgG ≥670 mg/dL. In addition, this study focused on nonpulmonary sepsis patients with serum IgG ≥670 mg/dL, analyzed the correlation between BALF (bronchoalveolar lavage fluid) IgA and lung imaging exudation status and duration of mechanical ventilation, and used flow cytometry to characterize the proportion of IgA1 + cells within the BALF-recovered CD45 + cell population and their IgA1 fluorescence intensity. Results There was no significant difference in serum immunoglobulin levels among patients with sepsis with different disease prognoses. In patients with serum IgG ≥670 mg/dL, each 100 mg/dL increase in serum IgA was associated with mortality after adjustment for age, sex, and SOFA score in the clinical cohort (adjusted HR, 1.269; 95% CI, 1.046–1.540; P = 0.016). The corresponding association did not reach statistical significance in MIMIC-IV (adjusted HR, 1.043; 95% CI, 0.999–1.089; P = 0.053). In patients with nonpulmonary sepsis, serum IgA concentrations and BALF IgA measurements were positively correlated with the change in pulmonary CT scores (Pearson's r = 0.5383; 95% CI, 0.2907–0.7183; and r = 0.5713; 95% CI, 0.2989–0.7576, respectively). Among patients with nonpulmonary sepsis, no statistically significant differences were observed between survivors and nonsurvivors in the proportion of CD19 + IgA1 + B cells among BALF-recovered CD45 + cells or in IgA1 fluorescence intensity within the IgA1 + B-cell population. Conclusions Serum IgA was associated with mortality after adjustment for age, sex, and disease severity in the clinical subgroup with IgG ≥670 mg/dL, but the adjusted MIMIC-IV analyses did not consistently support this association. Exploratory serum and BALF IgA measurements were associated with pulmonary CT abnormalities. These findings do not establish individual-level predictive performance or a causal role for IgA in pulmonary injury. Trial registration [ https://www.chictr.org.cn ] CHICTR, ChiCTR2400083751. Registered 30 April 2024.