Efficacy and safety of geptanolimab (GB226) for relapsed/refractory primary mediastinal large B-cell lymphoma: an open-label phase II study (Gxplore-003).

BACKGROUND: This study aimed to assess the efficacy and safety of geptanolimab (GB226), a fully humanized, recombinant anti-programmed cell death-1 monoclonal antibody, in Chinese patients with refractory or relapsed (r/r) primary mediastinal large B-cell lymphoma (PMBCL). METHODS: This was a multicenter, open-label, single-arm phase II study (Gxplore-003), conducted at 43 hospitals in China (NCT03639181). Patients received geptanolimab intravenously at a dose of 3 mg/kg every 2 weeks until documented confirmed disease progression, intolerable toxicity, or any other cessation criteria was met. The primary endpoint was objective response rate (ORR) in the full analysis set assessed by the independent review committee (IRC) according to the Lugano Classification 2014. RESULTS: This study was prematurely terminated due to the slow rate of patient accrual. Between Oct 15th, 2018 and Oct 7th, 2020, 25 patients were enrolled and treated. By the data cutoff date on Dec 23rd, 2020, the IRC-assessed ORR was 68.0% (17/25; 95% confidence interval [CI] 46.5-85.1%), with the complete response rate of 24%. The disease control rate was 88% (22/25; 95%CI 68.8-97.5%). Median duration of response was not reached (NR) (95%CI, 5.62 months to NR), with 79.5% of patients having response durations of more than 12 months. Median progression-free survival was NR (95%CI, 6.83 months to NR). Treatment-related adverse events (TRAEs) were reported in 20 of 25 (80.0%) patients, and grade 3 or higher TRAEs occurred in 11 of 25 (44%) patients. No treatment-related deaths occurred. The immune-related adverse events (irAEs) of any grade were observed in 6 (24.0%) patients, and no grade 4 or grade 5 irAEs were reported. CONCLUSION: Geptanolimab (GB226) demonstrated promising efficacy and a manageable safety profile in Chinese patients with r/r PMBCL.

Gene expression prediction based on neighbour connection neural network utilizing gene interaction graphs.

Having observed that gene expressions have a correlation, the Library of Integrated Network-based Cell-Signature program selects 1000 landmark genes to predict the remaining gene expression value. Further works have improved the prediction result by using deep learning models. However, these models ignore the latent structure of genes, limiting the accuracy of the experimental results. We therefore propose a novel neural network named Neighbour Connection Neural Network(NCNN) to utilize the gene interaction graph information. Comparing to the popular GCN model, our model incorperates the graph information in a better manner. We validate our model under two different settings and show that our model promotes prediction accuracy comparing to the other models.