Loss of CD20 expression as a mechanism of resistance to mosunetuzumab in relapsed/refractory B-cell lymphomas

Abstract

CD20 is an established therapeutic target in B-cell malignancies. The CD20 × CD3 bispecific antibody mosunetuzumab has significant efficacy in B-cell non-Hodgkin lymphomas (NHLs). Because target antigen loss is a recognized mechanism of resistance, we evaluated CD20 expression relative to clinical response in patients with relapsed and/or refractory NHL in the phase 1/2 GO29781 trial investigating mosunetuzumab monotherapy. CD20 was studied using immunohistochemistry (IHC), RNA sequencing, and whole-exome sequencing performed centrally in biopsy specimens collected before treatment at predose, during treatment, or upon progression. Before treatment, most patients exhibited a high proportion of tumor cells expressing CD20; however, in 16 of 293 patients (5.5%) the proportion was <10%. Analyses of paired biopsy specimens from patients on treatment revealed that CD20 levels were maintained in 29 of 30 patients (97%) vs at progression, where CD20 loss was observed in 11 of 32 patients (34%). Reduced transcription or acquisition of truncating mutations explained most but not all cases of CD20 loss. In vitro modeling confirmed the effects of CD20 variants identified in clinical samples on reduction of CD20 expression and missense mutations in the extracellular domain that could block mosunetuzumab binding. This study expands the knowledge about the occurrence of target antigen loss after anti-CD20 therapeutics to include CD20-targeting bispecific antibodies and elucidates mechanisms of reduced CD20 expression at disease progression that may be generalizable to other anti-CD20 targeting agents. These results also confirm the utility of readily available IHC staining for CD20 as a tool to inform clinical decisions. This trial was registered at www.ClinicalTrials.gov as #NCT02500407.

Graphical abstract

Figure 1.

Characterization of CD20 status in pre-mosunetuzumab tumor biopsy specimens by IHC. (A) Comparison of CD20 status determined by local CD20 staining per pathology reports vs central assessments using the dual CD20⁺PAX5⁺ IHC assay; shaded cells represent concordance between IHC assessments. (B) Percent CD20⁺PAX5⁺/PAX5⁺ cells in the tumor area before mosunetuzumab across NHL histologies from the GO29781 study. (C) CD20⁺PAX5⁺ staining relative to best overall response determined by investigator-assessed PET/CT imaging using the International Harmonization Project response criteria. CT, computed tomography; ND, not determined; NE, not estimable; PD, progressive disease; PET, positron emission tomography; PR, partial response; SD, stable disease.

Figure 2.

Changes in CD20 status after treatment with mosunetuzumab detected by CD20⁺PAX5⁺ dual IHC. (A) Changes in the proportion of CD20⁺PAX5⁺ cells in serial biopsy specimens collected on treatment. (B) CD20 status assessed in serial biopsy specimens collected at or after progression based on progression event. (C) Time to response in paired biopsy samples by CD20 status before mosunetuzumab or at progression; numbers in each row represent percent of CD20. Symbols indicating visits are colored black or pink to represent CD20 status at that visit: black (negative), pink (positive).

Figure 3.

CD20 variants identified by WES before mosunetuzumab and at progression affect CD20 status. (A) Mutations identified in tumor biopsies from patients treated with mosunetuzumab; missense (light blue), frameshift (dark blue), nonsense (red), and splice variants (purple). Predicted epitope for rituximab and obinutuzumab binding is bolded. (B) CD20 levels assessed as the percent CD20⁺ cells/total nucleated cells relative to PAX5⁺ cells/total nucleated cells for all biopsy specimens sequenced. Dots represent single samples colored by position and type of variant. EC, extracellular; IC, intracellular; SS, splice site; TM, transmembrane; TR, truncating. Panel A is adapted from Rushton et al.

Figure 4.

MS4A1 (the gene that encodes CD20) transcript profiling. (A) MS4A1 RNAseq-based expression relative to CD20⁺PAX5⁺ protein IHC staining in pre-mosunetuzumab tumor biopsies, and (B) in posttreatment tumor biopsies obtained on treatment (square) vs at progression (triangle). Loss of function (LoF) variants in CD20 identified by WES (red) and WES confirmed CD20 WT (purple). (C) Changes in the levels of MS4A1 in paired tumor biopsy specimens collected on treatment (left) or at progression (right) by best overall response (CR/PR, green; SD/PD, red). (D) Summary samples with multiple CD20 biomarker measurements (IHC, RNASeq, and WES) in samples collected before mosunetuzumab (top) or at progression (ie, at progression and on treatment) (bottom). LoF, loss of function; N/A, not available; nRPM, normalized reads per million.

Figure 5.

In vitro assessment of the effects of MS4A1 variants on CD20 expression and CD20 × CD3 activity. (A) CD20 protein expression by western blot in the SU-DHL-16–engineered cell lines. (B) Detection of CD20 by flow cytometry in SU-DHL-16–engineered cell lines. Intracellular expression was detected after permeabilization using anti-CD20 antibody targeting the C-terminus (H-1, BD-561174), and extracellular expression was detected using anti-CD20 antibody targeting ECL2 (2H7, BD-555623). (C) Immunofluorescence detection of CD20 using an intracellular antibody (ABCAM-78237): red fluorescence = CD20; blue fluorescence = 4′,6-diamidino-2-phenylindole (DAPI). (D) (left) CD20 × CD3–directed CD8-dependent cell killing of SU-DHL-16–engineered cell lines after 48 hours of treatment assessed by spectral flow cytometry. (Right) CD20 × CD3–directed CD8⁺ T-cell activation after 48 hours of treatment. T-cell activation monitored by flow cytometry by detecting activation markers CD69 and CD25. Flow cytometry data are represented as mean ± SD of 3 replicates. EV, empty vector; CD20-CD3, proof-of-concept CD20 × CD3 bispecific molecule; MFI, median fluorescence intensity; N, non-targeting control/CD3 antibody; parent, parental cell line; SD, standard deviation.