通过 AI 引导的 PET/CT 总代谢肿瘤体积（TMTV）和 LDH 水平评估的基线肿瘤负荷预测 CAR-T 在侵袭性 B 细胞淋巴瘤中的疗效

Disease burden is a critical determinant of outcomes in CAR-T therapy for B-cell lymphomas, and one of the most widely used techniques for its assessment is Total Metabolic Tumor Volume (TMTV) measured via [ 18 F]FDG PET/CT. Biological parameters may further refine the risk profile.

We analyzed baseline [ 18 F]FDG PET/CT scans from 40 patients treated with CAR-T, using an AI-based automated segmentation algorithm to determine TMTV.

Our analysis identified that a baseline TMTV greater than 48. 4 cm and elevated LDH independently predicted progression-free survival (PFS) after CAR-T therapy (HR 4. 28, p = 0. 007, and HR 8. 20, p < 0. 001, respectively).

We then proposed a 0-to-2 risk score, assigning one point each for elevated TMTV and elevated LDH. All patients with a score of two experienced a PFS of less than 90 days following CAR-T infusion. Among the remaining patients, those with 0 points versus 1 point demonstrated a 3-month PFS of 100% versus 85%, a 6-month PFS of 92% versus 53%, and a 12-month PFS of 83% versus 53%, respectively.

Importantly, patients with high baseline TMTV who achieved a TMTV reduction to less than 1. 99 cm by day 30 had a PFS of 66%, significantly better compared to those who did not achieve this reduction. AI-guided TMTV assessment, combined with LDH levels, provides a rapid and sensitive method for risk stratification at the bedside, which could help optimize patient management prior to CAR-T therapy.