Opportunities to improve antibiotic stewardship, and identification of blood biomarkers associated with bacteremia following CAR-T cell therapy.

BACKGROUND

Cytokine release syndrome (CRS) is frequent after chimeric antigen receptor T-cell therapy (CAR-T). CRS and bacteremia share clinical features, making it difficult to distinguish between the two and deliver targeted treatment. As a result, most patients with CRS receive empiric broad-spectrum antibiotics that may adversely impact long-term outcomes.

OBJECTIVES

The objectives of this study were to identify blood biomarkers that distinguish between CRS and bacteremia in the first month after CAR-T. We also describe the utilization of empiric antibiotics to identify opportunities for improved antimicrobial stewardship.

STUDY DESIGN

We identified patients who received CAR-T for hematologic malignancies between July 2013 and April 2024. We calculated the cumulative incidences of CRS and bacteremia within 30 days post-CAR-T. In a matched case control analysis, where three patients with CRS and no bacteremia (controls) were matched to each patient with bacteremia (cases), we described the clinical characteristics of CRS, bacteremia, and antibiotic exposure. We then assessed the levels and kinetics of six biomarkers (C-reactive protein [CRP], interleukin-6, ferritin, fibrinogen, lactate dehydrogenase, and D-dimer) for their ability to discriminate between bacteremia and CRS using Mann-Whitney U tests and generalized estimating equation (GEE) models.

RESULTS

Among 694 CAR-T recipients, 517 (75%; 95% confidence interval [CI], 71-78%) developed CRS. Bacteremia occurred in 19 patients with a cumulative incidence of 2.7% (95% CI, 1.7-4.2%) within 30 days and 1.4% (95% CI, 0.7-2.6%) within 14 days of CAR-T. The median time from blood culture to positive result was one day (interquartile range [IQR], 1-1). Among the 57 matched controls, CRS occurred a median of four days (IQR, 1-5) post-CAR-T, all were hospitalized for fevers, and 92% of CRS events were treated with empiric antibiotics for a median of seven days (IQR, 5-8). As a result, for every patient with bacteremia treated with antibiotics within the first 14 days, an estimated 52 patients with CRS and no serious bacterial infection received empiric antibiotics. The levels and kinetics of biomarkers were similar among cases and controls within 14 days after CAR-T. Beyond day +14, CRP, D-dimer, and ferritin levels were significantly higher among patients with bacteremia and distinguished cases from controls with moderate discrimination (area under the curve of 0.77, 0.77, and 0.81, respectively). In addition, increasing CRP and fibrinogen ≥ 14 days post-CAR-T were significantly associated with bacteremia (odds ratio [OR], 1.46; 95% CI, 1.07-2.01 and OR, 4.32; 95% CI, 1.28-14.61 per 0.25 log increase in biomarker level per day, respectively).

CONCLUSION

We demonstrate that most CAR-T recipients who developed CRS received empiric broad-spectrum antibiotics, but bacteremia was rare. Although blood biomarkers were unable to distinguish between CRS and bacteremia early after CAR-T, higher CRP, D-dimer, and ferritin and rising levels of CRP and fibrinogen ≥14 days post-CAR-T were associated with bacteremia and can facilitate earlier targeted interventions. These data highlight opportunities for improved antibiotic stewardship in the context of CRS, which is critical given the association between broad-spectrum antibiotic exposure, gut microbiome dysbiosis, and worse outcomes after CAR-T.