Abstract
Background: Prolonged thrombocytopenia (PT) is a major toxicity following chimeric antigen receptor (CAR) T-cell therapy, increasing hemorrhagic risk and limiting subsequent treatments. The etiology is complex, extending beyond lymphodepletion to include a CAR-T-induced inflammatory state within the bone marrow. We hypothesize that sustained high levels of pro-inflammatory cytokines, particularly IFN-γ, suppress hematopoiesis and impair megakaryopoiesis. With no standard management guidelines, we sought to characterize the incidence of PT, identify risk factors, and perform a comparative analysis of thrombopoietin-receptor agonist (TPO-RA) interventions in a large cohort treated with diverse CAR-T products.
Methods: We retrospectively analyzed 79 consecutive patients with relapsed/refractory B-cell non-Hodgkin lymphoma or multiple myeloma who received autologous, allogeneic, or multi-target CAR-T therapy at our center between March 2024 and July 2025. PT was defined as Grade ≥3 thrombocytopenia (platelet count <50 × 10⁹/L) persisting beyond Day +30 post-infusion without disease relapse. Multivariate logistic regression was used to analyze risk factors. The efficacy of TPO-RA interventions—including monotherapy (Eltrombopag, Avatrombopag, Hetrombopag, Lusutrombopag, Romiplostim, or rhTPO) and combination strategies—was evaluated. The primary endpoint was durable platelet recovery, defined as a sustained count of ≥50 × 10⁹/L for ≥7 consecutive days without transfusion. Time to recovery was analyzed using Kaplan-Meier methods.
Results: PT occurred in 48 of 79 patients (60.8%), with 20 (41.7%) having Grade 3 and 15 (31.3%) Grade 4 thrombocytopenia. Multivariate analysis identified significant risk factors for PT: prior intensive chemotherapy (HR 1.83, p=0.003), Grade ≥2 cytokine release syndrome (HR 2.45, p<0.001), baseline thrombocytopenia (HR 1.91, p=0.002), and elevated pre-lymphodepletion ferritin (HR 1.78, p=0.005). Sustained elevation of serum IFN-γ beyond day +21 was strongly associated with PT (81.3% vs. 41.7%, p<0.001).
Of the 48 patients with PT, 41 received TPO-RAs. Among monotherapies, romiplostim demonstrated the fastest recovery (median time to recovery [TTR] 39 days; Day 60 response 83%). Avatrombopag and lusutrombopag showed comparable efficacy (median TTR ~48 days; Day 60 response ~75%), followed by hetrombopag (median TTR 54 days; Day 60 response 61%).
Combination strategies were superior to monotherapy. A sequential approach of rhTPO followed by an oral TPO-RA (n=10) achieved a median TTR of 32 days, while concurrent romiplostim plus lusutrombopag (n=6) was the most potent regimen, with a median TTR of 29 days and a 100% response rate by day 60. Both combinations yielded significantly faster recovery than oral TPO-RA monotherapy (p<0.05). Early initiation of any TPO-RA (<Day +40) was associated with faster recovery (median 31 vs. 58 days, p=0.001). Adverse events were rare; two thromboembolic events were noted in patients receiving rhTPO. Flow cytometry confirmed higher frequencies of IFN-γ+ T cells in PT patients (p<0.001), which normalized with effective therapy.
Conclusion: Prolonged, severe thrombocytopenia is a frequent, inflammation-driven complication of CAR-T therapy. Our real-world data demonstrate a clear efficacy hierarchy among TPO-RAs, with romiplostim being the most effective single agent. Crucially, combination TPO-RA strategies, particularly those involving romiplostim or sequential rhTPO, significantly accelerate platelet recovery in severe cases without increasing toxicity. These findings support early, intensified TPO-RA intervention for high-risk patients and underscore the need for prospective trials to optimize these strategies and explore novel agents that mitigate the underlying inflammatory bone marrow damage.
