Systemic Inflammation Mediates the Association Between Admission Hyperglycemia and Pulmonary Infection or Prognosis in Acute Ischemic Stroke.

Among 2233 AIS patients, 26.3% had persistent hyperglycemia and 13.6% developed stress-induced transient hyperglycemia.

• Patients were stratified into three groups: normoglycemia (NG, n=1341), persistent hyperglycemia (PHG, n=588), and stress-induced transient hyperglycemia (SIH, n=304).
• Data were collected from January 2018 to February 2024 with 12-month follow-up.
• PHG accounted for 26.3% and SIH accounted for 13.6% of the total enrolled population.

Rates of stroke-associated pneumonia and poor prognosis were highest in the SIH group, intermediate in the PHG group, and lowest in the NG group.

• Primary outcomes were in-hospital stroke-associated pneumonia (SAP) and poor functional outcome (mRS >2) at 12 months.
• The gradient pattern across groups (SIH > PHG > NG) was observed for both SAP incidence and poor prognosis rates.
• This ordering suggests stress-induced transient hyperglycemia carries worse clinical implications than persistent hyperglycemia.

Patients in the SIH group exhibited the highest systemic inflammatory levels across multiple markers.

• Inflammatory markers assessed included CRP, hsCRP, lnSII (log-transformed Systemic Immune-Inflammation Index), SIRI (Systemic Inflammation Response Index), and neutrophil-to-lymphocyte ratio (NLR).
• SII and SIRI were calculated from baseline blood cell counts.
• The SIH group had higher levels of all five inflammatory markers compared to both PHG and NG groups.

Both tertile 3 of lnSII and SIRI were significantly associated with higher risk of SAP independent of admission glycemia status.

• Confounders were identified using LASSO regression before adjustment.
• The association between highest tertile of lnSII and SIRI with SAP was independent of glycemic group assignment.
• Higher lnSII and SIRI were significantly associated with poor prognosis at 12 months only in the SIH group, not across all glycemia groups.

Mediation analysis demonstrated that lnSII partially mediated the association between glycemic status and clinical outcomes.

• The mediation proportion for lnSII was 16.7% for SAP and 10.8% for 12-month prognosis.
• SIRI showed a similar mediating effect to lnSII for both outcomes.
• Both inflammatory indices partially, but not fully, explained the relationship between hyperglycemia and clinical outcomes.

Prediction models incorporating clinical variables and inflammatory indices achieved high discriminative performance for both outcomes.

• The prediction model incorporating clinical variables and lnSII or SIRI yielded an AUC of approximately 0.90 for SAP.
• The same model yielded an AUC of approximately 0.84 for 12-month prognosis.
• Both lnSII and SIRI were used as model components yielding similar AUC values.