Background Chimeric antigen receptor (CAR) T cells are highly effective in depleting B cells and are therefore being tested for their safety and efficacy in autoimmune diseases. Cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS) are potential toxic effects of CAR T-cell therapy. We aimed to identify biomarkers predicting incidence and severity of CRS in CAR T-cell treated patients with autoimmune diseases. Methods This retrospective cohort study included patients with autoimmune diseases treated with autologous CAR T-cell therapy from six centres in Germany and one centre in China. Demographic and disease-specific characteristics were recorded. Patients from Germany were treated with CD19 targeting CAR T cells. Patients from China were treated with either dual target CD19–BCMA CAR T cells or co-infused with CD19 CAR T cells and BCMA CAR T cells. Baseline biomarkers (B-cell counts and concentrations of IL-6, C-reactive protein [CRP], and ferritin) were recorded at the eligibility assessment visit before study entry or named patient programme use. IL-6 was measured by ELISA. The primary outcomes were the occurrence of CRS and ICANS. The severity of CRS and ICANS was classified using the American Society for Transplantation and Cellular Therapy criteria. Multivariable logistic regression was used to define factors associated with CRS occurrence. There was no involvement of people with lived experience in study design. Findings Between March 1, 2021, and July 1, 2025, we included 108 patients with autoimmune diseases who received autologous CAR T-cell therapy. The median age at administration was 34 years (IQR 27–44). 76 (70%) of 108 patients were women and 32 (30%) were men. 86 (80%) of 108 patients developed CRS after CAR T-cell therapy (grade one in 71 [66%] patients, grade two in 14 [13%] patients, and grade three in one [1%] patient). ICANS occurred in three (3%) patients (grade one in two [2%] patients and grade two in one [1%] patient). Patients who developed CRS had significantly higher baseline B-cell counts (43 cells per μL [IQR 10–108] vs 6 cells per μL [0–18]; p=0·0002), IL-6 concentrations (10 pg/ml [IQR 4–62] vs 3 pg/ml [2–6]; p=0·0003), and CRP concentrations (5 mg/L [IQR 2–14] vs 3mg/L [1–5]; p=0·042) compared with patients without CRS. Multivariable logistic regression confirmed the association between CRS, B-cell counts (OR 1·31 [95% CI 1·06–1·67]; p=0·012) and IL-6 concentrations (OR 2·28 [95% CI 1·30–4·65]; p=0·0001). Interpretation Higher baseline B-cell counts and systemic inflammation are associated with the development of CRS after CAR T-cell therapy in patients with autoimmune diseases. These data might help to identify patients with autoimmune disease that are at higher risk of developing CRS after CAR T-cell therapy. Funding Deutsche Forschungsgemeinschaft, Collaborative Research Centers, Clinician Scientist Program NOTICE.
BACKGROUND:Outcome measures for lupus arthritis that are not sensitive to change might contribute to misleading results in systemic lupus erythematosus (SLE) trials. We aimed to optimise lupus arthritis assessment by developing a novel composite disease activity measure, derived against ultrasound synovitis and incorporating participant-reported outcomes. METHODS:In this cohort study we derived the Lupus Arthritis and Musculoskeletal Disease Activity (LAMDA) score from baseline data in a prospective, longitudinal, multicentre study (USEFUL) of participants with SLE receiving intramuscular glucocorticoid therapy for lupus arthritis recruited from seven hospitals in England. Penalised multiple quantile (median) regression of total combined grey scale and power Doppler scores from bilateral hand and wrist joint ultrasounds on inflammatory arthritis core set variables defined by the American College of Rheumatology (tender joint count in 68 joints, swollen joint count in 66 joints, Health Assessment Questionnaire Disability Index, erythrocyte sedimentation rate [ESR], and visual analogue scales [VAS] for patient's musculoskeletal pain and disease activity and physician's musculoskeletal disease activity), and early morning stiffness severity VAS was used to derive the score. We assessed convergent construct validity, known groups validity, and responsiveness. We established thresholds of meaning for trial entry, minimal disease activity, participant acceptable symptom state, and minimal clinically important improvement using receiver operating characteristic curve analysis. A separate external validation cohort was recruited from Leeds Teaching Hospitals NHS Trust, clinical outcomes and treatment intention were collected by clinicians who were not involved in the USEFUL study. The construct validity of LAMDA within this external validation cohort was assessed, including discrimination of treatment intention and participant acceptable symptom state. People with lived experience of SLE were involved in the design and delivery of both the USEFUL study and the present study. FINDINGS:133 participants recruited between Oct 20, 2016, and Dec 20, 2018, were included from the USEFUL study; mean age was 47·0 years (IQR 35·0-55·0), 126 (95%) were female, seven (5%) were male, and 82 (62%) were White. The model retained four variables in the LAMDA score: swollen joint count in 66 joints, ESR, VAS for physician musculoskeletal disease activity, and participant musculoskeletal pain. High intraclass correlation (>0·99) supported directly substituting swollen joint count in 28 joints for swollen joint count in 66 joints to improve feasibility. LAMDA showed good construct validity, correlating with conventional disease activity measures, ultrasound findings, and patient-reported outcomes. LAMDA was responsive, yielding a medium-to-large early treatment effect following glucocorticoid therapy (effect size 0·40, p<0·0001). Provisional thresholds of meaning showed good logical consistency with participant characteristics and patient-reported change in musculoskeletal pain. The external validation cohort included 44 participants, recruited between Nov 15, 2017, and July 23, 2020; median age was 48·5 years (IQR 39·0-56·0), 38 (86%) were female, six (14%) were male, and 32 (73%) were White. The findings from the external validation cohort were supportive of the initial findings. LAMDA was correlated with multiple participant reported outcome measures and the proposed thresholds of meaning outperformed swollen joint counts alone in discriminating treatment intention and participant acceptable symptom state. INTERPRETATION:LAMDA is a novel ultrasound-derived clinical disease activity measure that should improve the assessment of lupus arthritis in clinical trials and real-world practice. Further validation is underway in large multinational randomised controlled trials. FUNDING:None.
