The CAR-HEMATOTOX risk-stratifies patients for severe infections and disease progression after CD19 CAR-T in R/R LBCL

Abstract

Background: CD19-directed chimeric antigen receptor T-cell therapy (CAR-T) represents a promising treatment modality for an increasing number of B-cell malignancies. However, prolonged cytopenias and infections substantially contribute to the toxicity burden of CAR-T. The recently developed CAR-HEMATOTOX (HT) score-composed of five pre-lymphodepletion variables (eg, absolute neutrophil count, platelet count, hemoglobin, C-reactive protein, ferritin)-enables risk stratification of hematological toxicity.

Methods: In this multicenter retrospective analysis, we characterized early infection events (days 0-90) and clinical outcomes in 248 patients receiving standard-of-care CD19 CAR-T for relapsed/refractory large B-cell lymphoma. This included a derivation cohort (cohort A, 179 patients) and a second independent validation cohort (cohort B, 69 patients). Cumulative incidence curves were calculated for all-grade, grade ≥3, and specific infection subtypes. Clinical outcomes were studied via Kaplan-Meier estimates.

Results: In a multivariate analysis adjusted for other baseline features, the HT score identified patients at high risk for severe infections (adjusted HR 6.4, 95% CI 3.1 to 13.1). HT^high patients more frequently developed severe infections (40% vs 8%, p<0.0001)-particularly severe bacterial infections (27% vs 0.9%, p<0.0001). Additionally, multivariate analysis of post-CAR-T factors revealed that infection risk was increased by prolonged neutropenia (≥14 days) and corticosteroid use (≥9 days), and decreased with fluoroquinolone prophylaxis. Antibacterial prophylaxis significantly reduced the likelihood of severe bacterial infections in HT^high (16% vs 46%, p<0.001), but not HT^low patients (0% vs 2%, p=n.s.). Collectively, HT^high patients experienced worse median progression-free (3.4 vs 12.6 months) and overall survival (9.1 months vs not-reached), and were hospitalized longer (median 20 vs 16 days). Severe infections represented the most common cause of non-relapse mortality after CAR-T and were associated with poor survival outcomes. A trend toward increased non-relapse mortality in HT^high patients was observed (8.0% vs 3.7%, p=0.09).

Conclusions: These data demonstrate the utility of the HT score to risk-stratify patients for infectious complications and poor survival outcomes prior to CD19 CAR-T. High-risk patients likely benefit from anti-infective prophylaxis and should be closely monitored for potential infections and relapse.

Keywords: hematologic neoplasms; receptors, chimeric antigen; translational medical research.

Figure 1

Post-CAR-T infectious complications are common in the real-world setting. (A) Cohort description: Cohort A includes 179 patients derived from the publication describing the CAR-HEMATOTOX model (Hamburg and Lyon did not participate in this study). Cohort B represents an independent validation cohort of 69 patients. Infections were graded on a five-grade scale with grade ≥3 infections indicating severe infections. (B) Overview of fever syndromes occurring within the first 10 days after CAR-T. (C) The maximum infection grade is depicted in the 112 patients (45%) that developed an infectious complication. The y-axis indicates the absolute number of patients that developed the respective infection grade, while the x-axis describes the infection grade (grades 1–5). The relative distribution of infection grades in percent is superimposed on the top of the bars. (D) Clinical source of infection of the 193 infection events. (E) Infection densities per 100 patient days are displayed for fungal, viral, bacterial and all subtypes of infections occurring during the first 90 days after CAR-T. (F, G) Cumulative incidence curves describing the time-to-first-infection in the first 90 days after CAR T for all-grade (F) and severe (grade ≥3) (G) infections. Cumulative incidences are depicted for all infections (gray), as well as in cohort A (dark blue, n=179) and cohort B (brown, n=69). The p value by Mantel-Cox log-rank test comparing cohorts A vs B is depicted on the graph inset. CAR-T, chimeric antigen receptor T-cells; CRP, C-reactive protein; GI, gastrointestinal; NNPF, non-neutropenic fever; NPF, neutropenic fever; CNS, central nervous cystem; HT, CAR-HEMATOTOX.

Figure 2

A high CAR-HEMATOTOX score results in a higher incidence and severity of post-CAR-T infections. (A) Relative distribution of infection grades for all infection subtypes comparing HT high versus low patients. Infection grades (1–5°) are color-coded in shades of green with the connecting green and gray lines and percentage numbers comparing all-grade and grade ≥3 infections, respectively, in HT high versus low patients. Significance values were determined by Fisher’s exact test (*p<0.05, **p<0.01, ***p<0.001, ****p<0.0001). (B) Infection densities per 100 patient days for the first 90 days after CAR-T by HT score. (C–G) Cumulative incidence curves (D0-90) by HT score for any-grade (C), grade ≥3 (D), and bacterial (E) infections. HRs adjusted for other host factors were calculated by Cox proportional-hazards model. (F) Relative distribution of infection grades for bacterial infections comparing HT low versus high patients. CAR-T, chimeric antigen receptor T-cells; HT, CAR-HEMATOTOX.

Figure 3

The CAR-HEMATOTOX, prolonged corticosteroid use and prolonged neutropenia represent independent risk factors for severe post-CAR-T infections. (A, B) Forest plots of the multivariate analysis performed as binary logistic regression analysis for the outcome of grade ≥3 infection. (A) Multivariate analysis of host factors determined prior to lymphodepletion (eg, age, disease entity, prior autologous stem cell transplantation, ECOG performance status, LDH >upper limit of normal, absolute lymphocyte count <200 /µL, immunoglobulin G levels <4 g/L, CAR-HEMATOTOX score ≥2). (B) Multivariate analysis of post-CAR-T factors (eg, CAR product—Tisa-cel, ICU admission, CRS/ICANS ASTCT grade ≥2°, tocilizumab administration, corticosteroid use ≥9 days, severe neutropenia ≥14 days, antibacterial prophylaxis use, G-CSF use). Adjusted p values accounting for the respective covariates are displayed on the graph inset. Variables reaching statistical significance (p<0.05) are highlighted in red (increased odds) or blue (decreased odds). (C, D) Binary logistic regression analysis for grade ≥3 infections for the time-dependent variables steroid use (C) and neutropenia (D). The cumulative duration of steroid use was calculated between days 0–21 (at least ≥10 mg dexamethasone equivalent per day). The duration of severe neutropenia (ANC<500 /µL) was determined between days 0–60. The p value is depicted for the Likelihood ratio test (G-squared) with light shading indicating the 95% asymptotic confidence bands. CAR-T, chimeric antigen receptor T-cells; CRS, cytokine release syndrome; ICANS, immune effector cell-associated neurotoxicity syndrome; LDH, lactate dehydrogenase; ANC, Absolute Neutrophil Count; ICU, Intensive Care Unit; ECOG, Eastern Cooperative Oncology Group; ASTCT, American Society for Transplantation and Cellular Therapy; G-CSF, Granulocyte-colony stimulating factor; PPX, prophylaxis.

Figure 4

HT high patients receiving antibacterial prophylaxis exhibit lower rates of bacterial infections. (A) Comparison of intravenous antibiotic use during the first 10 days after CAR-T transfusion in HT low (green) vs HT high (red) patients. Pip./Tazo=piperacillin/tazobactam, 3/4G = 3^rd or fourth generation cephalosporin. Only antibiotics that were employed at least 10 times were included, for detailed list see online supplemental table 5; p values determined by Fisher’s exact test. (B, C) Cumulative incidence curves (day 0–90) for bacterial infections by antibacterial prophylaxis use (PPX) in HT low (B) and HT high (C) patients. Light shading indicates patients receiving antibacterial prophylaxis, while dark shading denotes no antibacterial prophylaxis use; p values were determined by Mantel-Cox log-rank test comparing PPX vs no PPX. CAR-T, chimeric antigen receptor T-cells; HT, CAR-HEMATOTOX.

Figure 5

Patients with a high HT score and with severe infectious complications display poor survival outcomes. (A) 1-year non-relapse mortality (NRM) across all patients (gray) and in HT^high (red) vs HT^low (green) patients. The p value of the Mantel-Cox log-rank test comparing HT^high vs HT^low patients is depicted. (B) Median duration of hospitalization from the time point of lymphodepletion until the day of first discharge or death by HT score (data available for 246 patients). Statistical significance was determined by Mann-Whitney test. (C, D) Kaplan-Meier estimates of progression-free (C) and overall survival (D) by HT score. The median PFS and OS in months for HT^high (red) vs HT^low (green) patients is depicted on the lower left graph inset with the respective 95% CI. The HR with the 95% CI comparing HT^high (test) vs HT^high (comparator) in a univariate Cox regression model is depicted together with the p value of the log-rank test on the upper right graph inset. (E, F) Kaplan-Meier estimates of PFS (E) and OS (F) comparing patients with grade ≥3 infections (blue) to patients with grade <3 infections (orange). The p value of the log-rank test is depicted on the graph inset. OS, overall survival; PFS, progression-free survival; HT, CAR-HEMATOTOX.