The controversial interplay between hyperinflammation and immunosenescence in elderly critically ill co-infected...

Aging involves both immunosenescence and chronic low-grade inflammation, which can escalate into hyperinflammation during acute infection. However, how the opposing forces of immunosenescence and hyperinflammation jointly determine mortality, and whether viral-associated triple infection exacerbates this imbalance, remains poorly understood. This study investigates these two questions in critically ill co-infected patients by examining the interplay between hyperinflammation and lymphopenia in fatal outcomes.

This retrospective cohort study enrolled critically ill co-infected patients admitted hospital from February 2023 to September 2025, following rigorous inclusion and exclusion criteria to ensure data quality. Elastic net‑regularized logistic regression with 10‑fold stratified cross‑validation was applied to candidate variables prescreened by univariate logistic regression (p < 0.05); variables with non‑zero coefficients were then combined with the forced confounders (patient age and viral-associated triple infection) and refitted in a standard logistic regression model for unbiased estimation. Internal validation was then performed using 1000 bootstrap resamples. Model performance was evaluated via ROC curve, calibration curve, and predicted probability distribution plot. To better understand total protein (TP) and neutrophil-to-lymphocyte ratio (NLR) for mortality prediction, the individual TP, neutrophil, and lymphocyte counts of all patients were examined. To investigate whether viral-associated triple infection exacerbates immune dysfunction, we compared clinical and laboratory variables between patients with triple infection and those with non-viral double infection. Age-stratified analysis was also performed.

A total of 118 critically ill patients with bacterial-fungal co-infection were enrolled (mean age 68.6 years; 83% aged ≥ 60 years). Elastic net-regularized logistic regression (α = 0.3, λ = 0.4723) identified TP as an independent protective factor (OR = 0.876, P = 0.0014) against mortality, and NLR as an independent risk factor (OR = 1.026, P = 0.0182) for mortality. The final prediction model, Logit(P1) = 2.964 - 0.132×TP + 0.025×NLR + 0.035×patient age + 0.446×viral-associated triple infection, demonstrated promising discrimination (apparent AUC: 0.863 (95% CI: 0.786-0.94); optimism-corrected AUC: 0.836 (95% CI: 0.808-0.878); distribution overlap = 6.3%) and high calibration accuracy (Bootstrap-corrected calibration intercept=-0.176, slope = 0.801, mean absolute error = 0.0007). Among the 23 deceased patients, 87% had profound lymphopenia, while 74% exhibited neutrophilia. Additionally, 70% of patients had protein levels below 60 g/L, with 9 falling below 50 g/L. Compared to those non-viral double infection patients (n = 39), patients with viral-associated triple infection (n = 79) exhibited a more pronounced hyperinflammation-lymphopenia imbalance, with higher NLR, neutrophils, PCT, and CRP, along with lower lymphocytes, T-cell subsets, TP, and prealbumin. These findings were consistent in the age‑stratified subgroup of patients aged < 70 years (all FDR-adjusted q < 0.05).

This study reveals immune dysregulation characterized by hyperinflammation-immunosenescence coexistence; a preliminary prediction model (TP, NLR, age, viral coinfection) requires prospective multicenter validation.