Blockade of immune checkpoint pathways by programmed cell death protein 1 (PD-1) antibodies has demonstrated broad clinical efficacy against a variety of malignancies. Sintilimab, a highly selective, fully human monoclonal antibody (mAb), blocks the interaction of PD-1 and its ligands and has demonstrated clinical benefit in various clinical studies. Here, we evaluated the affinity of sintilimab to human PD-1 by surface plasmon resonance and mesoscale discovery and evaluated PD-1 receptor occupancy and anti-tumor efficacy of sintilimab in a humanized NOD/Shi-scid-IL2rgamma (null) (NOG) mouse model. We also assessed the receptor occupancy and immunogenicity of sintilimab from clinical studies in humans (9 patients with advanced solid tumor and 381 patients from 4 clinical studies, respectively). Sintilimab bound to human PD-1 with greater affinity than nivolumab (Opdivo®, MDX-1106) and pembrolizumab (Keytruda®, MK-3475). The high affinity of sintilimab is explained by its distinct structural binding mode to PD-1. The pharmacokinetic behavior of sintilimab did not show any significant differences compared to the other two anti-PD-1 mAbs. In the humanized NOG mouse model, sintilimab showed superior PD-1 occupancy on circulating T cells and a stronger anti-tumor effect against NCI-H292 tumors. The strong anti-tumor response correlated with increased interferon-γ-secreting, tumor-specific CD8+ T cells, but not with CD4+ Tregs in tumor tissue. Pharmacodynamics testing indicated a sustained mean occupancy of ≥95% of PD-1 molecules on circulating T cells in patients following sintilimab infusion, regardless of infusion dose. Sintilimab infusion was associated with 0.52% (2/381 patients) of anti-drug antibodies and 0.26% (1/381 patients) neutralizing antibodies. These data validate sintilimab as a novel, safe, and efficacious anti-PD-1 mAb for cancer immunotherapy.

Background: Programmed cell death 1 (PD-1) is an inhibitory immune checkpoint expressed on activatedT cells. Upon the formation of T cell receptor (TCR)-pMHC complexes, concomitant PD-1 ligation to its ligands programmed death-ligand 1 (PD-L1) or programmed death-ligand 2 (PD-L2) downregulates TCR signaling and effector function. Here we describe the preclinical characterization of Sintilimab, a fully human IgG4 antibody that potently blocks PD-1 interactions with PD-L1 and PD-L2. Methods: The binding affinity and blockade function were detected by using surface plasmon resonance (SPR), Enzyme-linked immunosorbent assay (ELISA) and flow cytometry. The biology function properties were measured with luciferase assay and mixed lymphocyte reaction assay. In vivo anti-tumor function and preclinical pharmacokinetic (PK) were identified with human PD-1 transgenic mice and non-human primates separately. Results: Sintilimab can specifically and strongly bind to human PD-1 (hPD-1) and cynomolgus PD-1 and the affinity of Sintilimab to human PD-1 was measured at 0.3 nm via surface SPR, and displayed slow dissociation kinetics. Sintilimab can block the interaction of PD-1 to PD-L1 and PD-L2 and induce high secretion levels of interferon (IFN)-γ and interleukin (IL)-2 in primary T cell assays. In humanized hPD-1 knock-in mouse models, Sintilimab showed potent anti-tumor activity and increased tumor-infiltrating CD8/CD4 T cell and CD8/ Treg ratios. Preclinical experimentation in non-human primates following a single intravenous infusion of Sintilimab at 1, 6 and 30 mg/kg presented with no signs of drug-related toxicity, and showed typical PK characteristics of an IgG antibody. Conclusions: Sintilimab has desirable preclinical attributes that supports its clinical development for cancer treatment.

In biomedical data mining, the gene dimension is often much larger than the sample size. To solve this problem, we need to use a feature selection algorithm to select feature gene subsets with a strong correlation with phenotype to ensure the accuracy of subsequent analysis. This paper presents a new three-stage hybrid feature gene selection method, that combines a variance filter, extremely randomized tree, and whale optimization algorithm. First, a variance filter is used to reduce the dimension of the feature gene space, and an extremely randomized tree is used to further reduce the feature gene set. Finally, the whale optimization algorithm is used to select the optimal feature gene subset. We evaluate the proposed method with three different classifiers in seven published gene expression profile datasets and compare it with other advanced feature selection algorithms. The results show that the proposed method has significant advantages in a variety of evaluation indicators.