Persistent and pathogen-specific infection risk during long-term survivorship after CD19 chimeric antigen receptor T-cell therapy: an Australian multicentre cohort study.

OBJECTIVES

Infection remains the leading cause of nonrelapse mortality after CD19 chimeric antigen receptor T-cell (CAR-T) therapy. However, the aetiology and risk of late infections, particularly beyond 1 year after infusion, remain poorly defined.

METHODS

This national multicentre retrospective cohort study included adults receiving standard-of-care CD19 CAR-T for aggressive lymphoma across five Australian centres between 2019 and 2023, with follow-up to March 2025. Infections were classified as microbiologically confirmed or clinically defined using consensus definitions. Outcomes included infection incidence, timing, microbiological aetiology, and pathogen-specific risk factors. Cause-specific Cox proportional hazards regression, with death as a competing risk, evaluated risks for microbiologically confirmed infections and early (day 0-30), medium (day 31-365), and late (≥365 days) infection events, with Firth regression used for sensitivity analyses.

RESULTS

Among 291 patients, 49.5% (144/291) experienced ≥1 infection (249 events); 36% (106/291) developed a severe infection (grade ≥3 Common Terminology Criteria for Adverse Events [CTCAE] and 5% (16/291) suffered a fatal infection. Of 146 patients evaluable after 1 year (median follow-up, 600 days; interquartile range, 420-900), 40% experienced a late infection (57/146) and 18% (27/146) a severe late infection. Respiratory viral infections (RVIs) were the most common late infections (37/89 events), and the proportion of severe RVIs did not decline over time. Early infection (OR, 2.96; 95% CI, 1.32-6.87; p 0.009) was associated with infection between days 31 and 365. No predictors of infections beyond 1 year were identified. Fungal infections were uncommon (6% of cohort, 17/291) but showed a distinct risk profile. Cumulative dexamethasone exposure and antecedent bacterial infection were associated with invasive fungal disease in multivariable analysis (p <0.05). A threshold of ≥56 mg dexamethasone-equivalent within the first 28 days predicted fungal risk (95% CI, 1.6-27.1; area under the curve, 0.71; p 0.015).

CONCLUSIONS

Late infections were common and clinically significant, with RVIs remaining a major source of morbidity. Fungal infection risk was associated with corticosteroid exposure extending beyond 6 months, supporting ongoing infection surveillance and prevention strategies beyond the first postinfusion year.