Accounting for B-cell Behavior and Sampling Bias Predicts Anti-PD-L1 Response in Bladder Cancer

Ilya A. Dyugay, Daniil K. Lukyanov, Maria A. Turchaninova, Ekaterina O. Serebrovskaya, Ekaterina A. Bryushkova, Andrew R. Zaretsky, Oybek Khalmurzaev, Vsevolod B. Matveev, Mikhail Shugay, Pavel V. Shelyakin, Dmitriy M. Chudakov

Research output: Contribution to journalArticlepeer-review

2 Citations (Scopus)

Abstract

Cancer immunotherapy is predominantly based on T cell-centric approaches. At the same time, the adaptive immune response in the tumor environment also includes clonally produced immunoglobulins and clonal effector/memory B cells that participate in antigen-specific decisions through their interactions with T cells. Here, we investigated the role of infiltrating B cells in bladder cancer via patient dataset analysis of intratumoral immunoglobulin repertoires. We showed that the IgG1/IgA ratio is a prognostic indicator for several subtypes of bladder cancer and for the whole IMVigor210 anti-PD-L1 immunotherapy study cohort. A high IgG1/IgA ratio associated with the prominence of a cytotoxic gene signature, T-cell receptor signaling, and IL21-mediated signaling. Immunoglobulin repertoire analysis indicated that effector B-cell function, rather than clonally produced antibodies, was involved in antitumor responses. From the T-cell side, we normalized a cytotoxic signature against the extent of immune cell infiltration to neutralize the artificial sampling-based variability in immune gene expression. Resulting metrics reflected proportion of cytotoxic cells among tumor-infiltrating immune cells and improved prediction of anti-PD-L1 responses. At the same time, the IgG1/IgA ratio remained an independent prognostic factor. Integration of the B-cell, natural killer cell, and T-cell signatures allowed for the most accurate prediction of anti-PD-L1 therapy responses. On the basis of these findings, we developed a predictor called PRedIctive MolecUlar Signature (PRIMUS), which outperformed PD-L1 expression scores and known gene signatures. Overall, PRIMUS allows for reliable identification of responders among patients with muscle-invasive urothelial carcinoma, including the subcohort with the low-infiltrated "desert" tumor phenotype.

Original languageEnglish
Pages (from-to)343-353
Number of pages11
JournalCancer immunology research
Volume10
Issue number3
DOIs
Publication statusPublished - 1 Mar 2022

Fingerprint

Dive into the research topics of 'Accounting for B-cell Behavior and Sampling Bias Predicts Anti-PD-L1 Response in Bladder Cancer'. Together they form a unique fingerprint.

Cite this