Association of Tumor Mutational Burden and Immune Gene Expression with Response to PD-1 Blockade by Sasanlimab Across Tumor Types and Routes of Administration

Abstract

Background: Sasanlimab is a monoclonal antibody that binds to the programmed cell death receptor 1 (PD-1). Anti-PD-1 monoclonal antibodies have improved patient clinical outcomes; however, not all treated patients derive clinical benefit. Further insights on potential biomarkers beyond PD-L1 expression levels would help to identify the patients most likely to respond to treatment.

Objective: This study evaluated tumor biopsies from patients treated with intravenous or subcutaneous sasanlimab to identify biomarkers of response and characterize pharmacodynamic activity.

Methods: Anti-PD-1/PD-ligand 1 (PD-L1)-naive patients with advanced solid tumors received sasanlimab intravenously at 1, 3, or 10 mg/kg every 3 weeks (n = 23) or subcutaneously at 300 mg every 4 weeks (n = 15). Best tumor percentage change from baseline was determined by RECIST. Whole-exome DNA and RNA sequencing were performed in tumor samples collected from treated patients at protocol-defined timepoints. PD-L1 and CD8 protein expression were evaluated in tumor biopsies by immunohistochemistry. Associations with response were assessed by linear regression analysis.

Results: Baseline tumor mutational burden (TMB), as well as PD-L1 and CD8 expression, were significantly associated with response to sasanlimab across the multiple dose levels, routes of administration, and range of tumor types evaluated. TMB is an independent biomarker from the various tumor inflammatory genes and signatures evaluated. Gene set enrichment analysis showed that higher baseline expression levels of genes related to the interferon-γ and PD-1 signaling pathways and the cell cycle were significantly associated with response to sasanlimab across tumor types. No differences were observed between routes of administration with regard to response to sasanlimab for the biomarkers of interest (TMB, PD-L1, CD8, and interferon-γ signature). Evaluation of pharmacodynamic changes showed increased tumor expression of genes enriched in adaptive immune response pathways.

Conclusions: Our findings indicate an active, immunomodulatory mechanism for the anti-PD-1 antibody sasanlimab across different tumor types and routes of administration.

Trial registration: ClinicalTrials.gov identifier NCT02573259; registered October 2015.

Conflict of interest statement

S. Hu-Lieskovan disclosed consulting honoraria from Amgen, Genmab, Xencor, and Merck; research funding from Bristol Myers Squibb, Merck, and Vaccinex. F. Braiteh disclosed speaker bureau and advisory board honoraria from Pfizer. J. E. Grilley-Olson disclosed no relevant conflict of interest. X. Wang, A. Forgie, V. Bonato, I. A. Jacobs, and J. Chou were employees of and owned stock in Pfizer Inc. at the time of this study. ML Johnson disclosed institutional research funding from AbbVie, Adaptimmune, Amgen, Apexigen, Arcus Biosciences, Array BioPharma, Artios Pharma, AstraZeneca, ATRECA, BeiGene, BerGenBio, Birdie Pharmaceuticals/Seven and Eight Biopharmaceuticals, Boehringer Ingelheim, Calithera Biosciences, Checkpoint Therapeutics, Corvus, Curis, CytomX, Daiichi-Sankyo, Dracen Pharmaceuticals, Dynavax, EMD Serono, Genentech/Roche, Genmab, Genocea, GlaxoSmithKline, Gritstone Oncology, Harpoon Therapeutics, Hengrui Therapeutics, Immunocore, Incyte, Janssen, Lilly, Loxo Oncology, Lycera, Merck, Mirati Therapeutics, Neovia, Novartis, Pfizer, PMV Pharmaceuticals, Regeneron, Ribon Therapeutics, Sanofi, Shattuck Labs, Silicon Therapeutics, Stemcentrx, Syndax, Takeda, Tarveda Therapeutics, TCR2 Therapeutics, TMUNITY Therapeutics, University of Michigan, and WindMIL Therapeutics; spouse role as contract lobbyist for Astellas and Otsuka Pharmaceuticals; and consulting honoraria (to institution) from AbbVie, Amgen, AstraZeneca, Boehringer Ingelheim, Bristol Myers Squibb, Calithera, Celgene, Daiichi Sankyo, Editas Medicine, Eisai, EMD Serono, G1 Therapeutics, Genentech/Roche, GlaxoSmithKline, Gritstone Oncology, Ideaya Biosciences, Incyte, Janssen, Lilly, Loxo Oncology, Merck, Mirati Therapeutics, Novartis, Pfizer, Ribon Therapetics, Sanofi and WindMIL Therapeutics.

© 2021. Pfizer Inc.

Figures

Fig. 1

Flow of samples through the study. C2D8 cycle 2 day 8, IHC immunohistochemistry, IV intravenous, PD-L1 programmed cell death ligand 1, q3w every 3 weeks, q4w every 4 weeks, RNA-Seq RNA sequencing, SC subcutaneous

Fig. 2

Baseline TMB and response to treatment with sasanlimab (p = 0.0134). The statistics were calculated by linear regression of TMB on best percentage change from baseline in tumors. The blue solid line represents the regression model fit along with its 95% confidence intervals (dark gray shaded area). Dashed lines indicate the cutoff between PR and SD (30% tumor shrinkage) and the cutoff between SD and disease progression (20% tumor growth). Adj adjusted, PR partial response, SD stable disease, TMB tumor mutational burden

Fig. 3

Analysis of baseline PD-L1 (a–c) and CD8 (d, e) protein/RNA expression and response to treatment with sasanlimab. The statistics were calculated from linear regression of protein/RNA expression on best percentage change from baseline in IHC and RNA-Seq samples. Round dots: baseline tissues with both IHC and RNA-Seq profiling; triangles: baseline tissues with IHC only or RNA-Seq only. PD-L1 protein expression results presented as a percentage viable tumor cells expressing PD-L1 (membrane staining) (p = 0.805) and b percentage immune cells with PD-L1 expression/total immune cells (p = 0.258). c, e RNA-Seq analyses performed in non-separated, bulk RNA samples (p = 0.0294 and p = 0.0343, respectively). d CD8 protein expression evaluated by number of immune cells within the central tumor area (p = 0.79). The blue solid line represents the regression model fit along with its 95% confidence intervals (dark gray shaded area). Adj adjusted, IHC immunohistochemistry, PD-L1 programmed cell death ligand 1, RNA-Seq ribonucleic acid sequencing, TPM transcripts per million

Fig. 4

Analysis of genes associated with response to treatment with sasanlimab in baseline biopsies. a Genes positively and negatively associated with improved response (p < 0.05) are shown in green and red, respectively. b, c Dot plots display the association between CTLA4 or IFNG baseline gene expression and best percentage change from baseline in tumors (p < 0.0001 and p = 0.00271, respectively). p values for CTLA and IFNG were generated by limma [20]. d Dot plot of IFN-γ signature expression and best percentage change from baseline in tumors, analyzed by linear regression (p = 0.0201). The blue solid line represents the regression model fit along with its 95% confidence intervals (dark gray shaded area). IFN-γ signature expression was defined as the mean expression of 18 IFN-γ related genes (CD274, CXCR6, TIGIT, CD27, PDCD1LG2, LAG3, NKG7, PSMB10, CMKLR1, CD8A, IDO1, CCL5, CXCL9, HLA-DQA1, CD276, HLA-DRB1, STAT1, HLA-E). Adj adjusted, IFN-γ interferon-gamma, TPM transcripts per million

Fig. 5

Gene set enrichment analysis by pre-ranked GSEA on results from association analysis between baseline gene expression and response to treatment (Data4Cure with Reactome pathways). a Dots indicate Reactome pathways curated by Data4Cure and lines indicate the relation of the pathways. Network view’s colors indicate the significance of gene set enrichment: burgundy = pathways enriched in genes positively associated with response (good prognosis), green = pathways enriched in genes negatively associated with response (poor prognosis). b Most significantly enriched pathways; bar length shows the statistic from pre-ranked GSEA analysis, where larger absolute values indicate more significant enrichment. Dashed lines indicate q value = 0.05 [18]. APC/C anaphase-promoting complex/cyclosome, Cdc cell-division cycle protein, DNA deoxyribonucleic acid, ECM extracellular matrix, G1 gap 1, GSEA gene set enrichment analysis, HDAC histone deacetylase, M mitosis, PD-1 programmed cell death 1, S synthesis

Fig. 6

a Genes differentially expressed in on-treatment (C2D8) versus baseline biopsies. Up-regulated and down-regulated genes with p < 0.05 are shown in red and green, respectively. b Dot plots display expression of selected genes in on-treatment versus baseline biopsies: CXCL9 (p = 0.00168); PDCD1 (p = 0.00759); IFNG (p = 0.00727); CD274 (p = 0.211); CTLA4 (p = 0.421); CDKN1B (p = 0.0101). p values were generated by limma [20]. C2D8 Cycle 2, Day 8, CD274 cluster of differentiation 274, CDKN1B cyclin dependent kinase inhibitor 1B, CTLA4 cytotoxic T-lymphocyte-associated protein 4, CXCL9 C-X-C motif chemokine ligand 9, IFNG interferon gamma, PDCD1 programmed cell death protein 1, TPM transcripts per million

Fig. 7

Gene set enrichment analysis by pre-ranked GSEA on results from differential analysis between on-treatment and baseline biopsies (Data4Cure with Reactome pathways). a Network view of pathway expression analysis results using statistics from paired differential expression analysis. Significance of pathway enrichment: burgundy = pathways enriched in genes with increased expression after treatment, green = pathways enriched in genes with decreased expression after treatment. b Most significantly enriched pathways. Dashed lines indicate q value = 0.05 [18]. APOBEC3G apolipoprotein B mRNA editing enzyme-catalytic polypeptide-like 3G, AUF1 AU-rich element RNA-binding protein 1, Cdc cell-division cycle protein, CDK cyclin-dependent kinase, CENPA centromere protein A, COP1 constitutive photomorphogenic 1, DNA deoxyribonucleic acid, DVL disheveled proteins, ER endoplasmic reticulum, ERAD endoplasmic-reticulum-associated protein degradation, G1 gap 1, GSEA gene set enrichment analysis, Hh hedgehog, hnRNP heterogenous nuclear ribonucleoprotein, mRNA messenger ribobucleic acid, NF-κB nuclear factor kappa B, PAK p21-activated protein kinase, PD-1 programmed cell death 1, RNP ribonucleoprotein, S synthesis