BIOS-04. THERAPIES FOR LEPTOMENINGEAL DISEASE: A NETWORK META-ANALYSIS

Abstract INTRODUCTION Leptomeningeal disease (LMD) has limited head-to-head comparison of available therapies, including systemic chemotherapy (SysCT), intrathecal (IT) or intraventricular therapy (ITV), whole-brain radiotherapy (WBRT), or cranio-spinal irradiation (CSI). No network meta-analysis (NMA) of randomized controlled trials (RCTs) of LMD has been reported so far. METHODS NMA was conducted using published LMD RCTs and reported following PRISMA-NMA guidelines. Primary outcome was survival, assessed using hazard ratios (HRs) and 95% confidence intervals (95%CI). Survival times were converted to HRs under exponential distribution assumption. LogHR standard errors were calculated using deaths per RCT arm. P-scores were used to estimate probability of treatment being superior. Analysis was performed in R using frequentist approach, followed by back-transforming logHRs for indirect and direct comparisons. Incremental effects of individual treatment components were estimated using additive model. Heterogeneity and inconsistency were assessed using tau-squared (τ²), I-squared (I²), and Q-statistics. Analyses were performed separately for main subnetwork with disconnected subnetworks accommodated in additive model. RESULTS Seven studies were included, incorporating 7 pairwise comparisons and 9 treatments. 3 subnetworks were identified. Primary subnetwork analysis included 5 studies. Compared to SysCT±RT (reference), logHRs (95%CI) under random effects model were: IT/ITV DepoCyt±SysCT±RT -0.094(-0.468,0.281); IT/ITV methotrexate(MTX)±cytarabine(AraC)±SysCT±RT 0.8567(0.066,1.647). IT/ITV MTX±SysCT±RT 0.318(-0.163,0.798); ITV Thiotepa±SysCT±RT 0.438(-0.294,1.169). Heterogeneity and inconsistency were not detected (τ²=0,I²=0%,Q-statistic=0.57,P=0.45). P-scores were highest for IT/ITV DepoCyt±SysCT±RT (0.894), followed by SysCT±RT (0.769). Disconnected-component-NMA demonstrated incremental logHR: AraC 0.539(-0.347,1.425); IT/ITV DepoCyt 0.052(-0.566,0.670); IT/ITV MTX 0.179(-0.484,0.840); ITV thiotepa 0.298(-0.767,1.362); proton CSI -0.250(-0.726,0.225). No component demonstrated statistically significant incremental benefit, with moderate heterogeneity and adequate fit present (τ²=0.1017,I²=57%,Q-statistic=4.65,P=0.098). CONCLUSIONS IT/ITV DepoCyt±SysCT±RT and SysCT±RT were found most effective. Lack of significant incremental benefit warrants evaluating older therapies in contemporary RCTs.