The CAR-Hematotox Identifies Patients at High Risk for Prolonged Neutropenia, Infectious Complications and Prolonged Hospitalization Following CD19-CART in R/R LBCL

Introduction: Prolonged cytopenias and infectious complications substantially contribute to the toxicity burden of CD19-directed CAR T-cell therapy (CAR-T). The latter represent the most common cause of non-relapse mortality after CAR-T. We have recently described the CAR-HEMATOTOX (HT) score, which integrates parameters associated with pre-CAR-T hematopoietic reserve (e.g. ANC, hemoglobin, platelet count) and inflammation (e.g. CRP, ferritin) to predict CAR-T-related hematotoxicity (Rejeski et al, Blood 2021). Here, we describe the utility of the HT score for identifying patients at risk for post-CAR-T infectious complications.

Methods: In this multicenter retrospective analysis, we studied infection events occurring within the first 90 days after CAR-T in 248 patients receiving standard-of-care Axi-cel (n=138) or Tisa-cel (n=110) for relapsed/refractory LBCL. An episode of infection was defined as bacterial, viral or fungal based on microbiologic data or as a clinical syndrome of infection (e.g. pneumonia, cellulitis, cystitis) based on retrospective chart review. In the absence of clinicals signs or microbiologic data, fever was not counted as infection. Infection severity was classified on a 5-grade scale as mild, moderate, severe, life-threatening or fatal as previously established (Hill et al, Blood 2018). Severe (Grade ≥3) infections were defined as requiring i.v. anti-infective agents and/or hospitalization. Multivariate analysis was performed as binary logistic regression for the primary endpoint of grade ≥3 infection. Cumulative incidence curves were calculated as time-to-first-infection; hazard ratios were calculated by Cox proportional-hazards model.

Results: In our cohort, the median duration of severe neutropenia (ANC<500/µl) in the first 60 days after CAR-T was 9 days (95% CI: 8-10 days). HT ^High patients exhibited significantly prolonged neutropenia compared to their HT ^Low counterparts (13 vs. 6 days, p<0.001). In the first 10 days after CAR-T, 85% of patients developed either neutropenic (71%) or non-neutropenic fever (14%). The cumulative 90-day-incidence of infection was 43% overall and 24% for grade ≥3 infections. The large majority (>85%) of infections occurred within the first 30 days with a median time to first infection of 7 days (95% CI: 5-8 days).

Next, we performed multivariate analysis to study the influence of pre-CAR-T parameters on the binary outcome of grade ≥3 infection. When accounting for other host factors (e.g. sex, age, ASCT, LDH, disease entity), a high HT score was associated with an increased probability for grade ≥3 infection (Fig. 1A, OR = 7.6, 95% CI: 3.4 - 16.3). Overall, the rate of severe infections was significantly higher in HT ^High patients (38% vs. 8%, p < 0.001). When studying the distribution of infection grades, the majority of infections in HT ^Low patients were mild or moderate (83%). On the other hand, HT ^High patients predominantly developed grade ≥3 infections (58%) - including 17 life-threatening and 6 fatal infections in the first 90 days. The cumulative incidence of severe infections over time was significantly higher in HT ^High patients (Fig. 1B, HR = 6.4, 95% CI: 3.1 - 13.1). Of the infection subtypes, the negative impact of a high HT score was particularly evident for bacterial infections (HR = 2.2, 95% CI: 1.4 - 3.9).

This translated into a longer median duration of hospitalization in HT ^High patients (20 vs. 16 days, p < 0.001, updated analysis), highlighting the clinical relevance and potential health economic impact of the score. Non-relapse mortality in our cohort was 6.0% (15/248) and was predominantly attributed to an infectious cause (10 infections, 3 CRS, 1 ICANS, 1 accident). Notably, a high HT score was associated with inferior overall survival (median: 8.6 months vs. not-reached, p < 0.001, updated analysis).

Conclusions: These findings emphasize that early post-CAR-T infections are common in the real-world setting and are likely facilitated by prolonged neutropenia. The CAR-HEMATOTOX represents a useful and easy-to-apply clinical score to risk-stratify patients for these toxicities prior to lymphodepletion. The state of inflammation and hematologic reserve of the host thus predisposes for later infectious complications. In lieu of the high incidence of grade ≥3 infections in HT ^High patients, prospective evaluation of HT-adapted anti-infective prophylaxis and/or early G-CSF support appears warranted.

Figure 1

View largeDownload slide

Figure 1

View largeDownload slide

Close modal