Exploration of Immune-Modulatory Effects of Amivantamab in Combination with Pembrolizumab in Lung and Head and Neck Squamous Cell Carcinoma

Abstract

Immune checkpoint inhibitors are effective first-line therapy for solid cancers. However, low response rate and acquired resistance over time has led to the need for additional therapeutic options. Here, we evaluated synergistic antitumor efficacy of EGFR × MET targeting bispecific antibody, amivantamab with PD-L1 immunotherapy, pembrolizumab in head and neck squamous cell carcinoma (HNSCC) and lung squamous cell carcinoma tumor-bearing humanized patient-derived xenograft (PDX) models. We demonstrated that pembrolizumab or amivantamab alone was ineffective and that combination treatment induced a significant reduction of tumor growth in both models (P < 0.0001 and P < 0.01, respectively). It appeared that combination of amivantamab and pembrolizumab significantly enhanced infiltration of granzyme B-producing CD8 T cells was in the TME of HNSCC PDX (P < 0.01) and enhanced neoantigen-associated central memory CD8 T cells in circulating immune cells. Analysis of single-cell RNA transcriptomics suggested that the tumor cells dramatically upregulated EGFR and MET in response to PD-L1 immunotherapy, potentially creating a metabolic state fit for tumor persistence in the tumor microenvironment (TME) and rendered pembrolizumab ineffective. We demonstrated that EGFRHIGHMETHIGH subcluster displayed an increased expression of genes implicated in production of lactate [SLC16A3 and lactate dehydrogenase A (LDHA)] compared to the EGFRLOWMETLOW cluster. Accumulation of lactate in the TME has been associated with immunosuppression by hindering the infiltration of tumor killing CD8 T and NK cells. This study proved that amivantamab reduced glycolytic markers in the EGFRHIGHMETHIGH subcluster including SLC16A3 and LDHA and highlighted remodeling of the TME by combination treatment, providing rationale for additional therapy of amivantamab with PD-1 immunotherapy.

Significance: Amivantamab in synergy with pembrolizumab effectively eradicated EGFRHIGHMETHIGH tumor subcluster in the tumor microenvironment of head and neck squamous cell carcinoma and overcame resistance against anti-PD-1 immunotherapy.

Figure 1

Antitumor effects of amivantamab w/wo pembrolizumab in HNSCC and LUSC tumor–bearing humanized PDX preclinical models. A, The intensity of EGFR and MET in YHIM-3003 (HNSCC) and YHIM-2010 (LUSC) tumors. B, The tumor progression of YHIM-3003 model over 19 days showing significant tumor regression by the combination treatment of amivantamab (30 mpk) and pembrolizumab (10 mpk) compared to single treatment of amivantamab and pembrolizumab (P < 0.001, n = 10 in each group). C, Tumor growth inhibition represented in a waterfall plot of HNSCC PDX model at day 19. D, Survival curve of HNSCC PDX demonstrating improved survival by the combination treatment (n = 10 in each group). E, Treatment of LUSC PDX using amivantamab at 30 mpk in combination with pembrolizumab (10 mpk, n = 10 in each group). No rebound of tumor growth was observed after the termination of amivantamab and combination treatment. F, Tumor regression in LUSC PDX showing significant tumor reduction by combination treatment of amivantamab (10 mpk) and pembrolizumab (10 mpk) compared to single treatment of amivantamab (P < 0.05) and pembrolizumab (P < 0.001, n = 10 in each group). G, Survival curve of LUSC PDX demonstrating improved survival by the combination treatment.

Figure 2

Multiplex IHC showing T cell subpopulations in the tumor microenvironment of HNSCC PDX (YHIM-3003) tumor after combination treatment of amivantamab and pembrolizumab. A, Whole slide images of tumor scan showing segmentation of tissues and cells. Helper CD4 and cytotoxic CD8 T cells were marked with FoxP3 (regulatory T cells) and GZMB (active cytotoxic T cells). Cancer cells were stained with Pan-CK. B, Stained whole slide images were quantified and presented in bar plots showing GZMB⁺ CD8 T cells and regulatory T cells in the tumor microenvironment, tumor nest and stroma in different treatment groups. Proportion of GZMB⁺ CD8 T cells in the total TME and the tumor nest was significantly increased in the combination group (P < 0.05). Each bar in the bar plot represents five mice that were internally sacrificed.

Figure 3

Flow cytometry analysis of memory subsets of T cells in the different treatment groups of humanized HNSCC (YHIM-3003) and LUSC (YHIM-2010) PDX tumor of internally sacrificed mice (n = 5 in each treatment group). A, Heatmap of memory T cell subsets (central memory, effector memory, and effector T cells) and activation markers in the tumor samples of HNSCC PDX. B, Heatmap of memory T-cell subsets (central memory, effector memory, and effector T cells) and activation markers in the tumor samples of YHIM-2010. C, Factors that combination of amivantamab and pembrolizumab positively affected in each humanized PDX model and both models shared enhancement of CD8⁺ T central memory subset by combination therapy. D, Tumor reactive (CEA-stained) CD8 T cells in HNSCC PDX tumor were abundant in the combination treatment group and were significantly higher in proportion compared to the control group (8.28 ± 2.67 and 3.02 ± 0.75, respectively, P < 0.05).

Figure 4

Single-cell RNA sequencing analysis of EGFR^HIGH/MET^HIGH and EGFR^LOW/MET^LOW subclusters in the tumor of humanized HNSCC (YHIM-3003) PDX mice showing relative increase in EGFR and MET expressing subcluster after treatment of pembrolizumab (n = 5 in each treatment group). Top 50 genes were analyzed by log₂ fold change (FC) against P-values and immune related genes were divided into three categories as following: immunomodulation, metastasis potential and cancer progression (drug resistance/cancer stemness). A, Heatmap of tumor indicating a region of tumor subcluster with elevated expression of both EGFR and MET. B, Density plot illustrating EGFR^HIGH/MET^HIGH and EGFR^LOW/MET^LOW subclusters in treatment groups. Expression of EGFR and MET was relatively higher in the pembrolizumab treated group compared to the other treatment groups (C). D, Tumor subcluster with elevated dual expression of EGFR and MET (EM^HIGH) was define and analyzed for DEGs. E, EGFR and MET in EM^HIGH and EM^LOW, showing increased expression of both markers in EM^HIGH tumor cluster. F, DEG analysis of the top genes in the EGFR^HIGH/MET^HIGH tumor subcluster showing genes related to immunomodulation, tumor metastasis, drug resistance and cancer stemness in volcano plot. G, Multiplex IHC of the tumor tissue showing reduced expression of EGFR in tumor treated with combination treatment. EGFR+ site, colored green, represents tumor regions that were stained positive for EGFR and does not directly translate to the level of expression. Tumor with damaged or indistinct regions of tumor nest was treated as an outlier and was removed from each group. H, Fluorescence intensity of EGFR in randomized region of interest sites converted into average H-scores. H-scores ranged from 0 (H-0, blue) to 3 (H-3, red), with H-3 representing the highest intensity as shown in the top. Average H-score of EGFR expression in pembrolizumab treated mice was significantly higher than the other groups (P < 0.01).

Figure 5

LDHA and SLC16A3 are important regulators of glycolytic pathways and were upregulated in EGFR^HIGH/MET^HIGH tumor subcluster of HNSCC PDX (YHIM-3003). A, Correlation between expression of EGFR and biomarkers based on TCGA database. LDHA and SLC16A3 expression was positively correlated with expression of EGFR in both HNSCC and LUSC. B, Expression of LDHA and SLC16A3 was significantly increased in EGFR^HIGH/MET^HIGH tumor subcluster (EM^HIGH) of HNSCC PDX (left). Additionally, expression of LDHA and SLC16A3 increased in pembrolizumab treated group. C, Regulators of glycolysis (HK2, GPI, ALDOA, PGK1, PGAM1, ENO1, ENO2) comparatively increased in the EM^HIGH tumor subcluster (left) and pembrolizumab treated group (right). D, Regulators of hypoxia (HIF1A, HDAC1, KDM1A, KDM2A) and downstream signaling markers (CA9, VEGFA, TWIST1) increased in the EM^HIGH tumor subcluster (left) and pembrolizumab treated group (right). E, Total protein and surface expression of EGFR and MET demonstrated strong correlation in HNSCC and LUSC cell lines (left). H1703, LUSC human cancer cell, was treated with IFNγ for 24 hours to mimic the physiological response of pembrolizumab in the TME (right). Induction of IFNγ, though not significant, upregulated the expression of pEGFR, while pMET level significantly increased (P < 0.01). F, Protein expression of EGFR/p-EGFR, MET/p-MET, MCT4 (SLC16A3), and LDHA after 72 hours of amivantamab at 10 mg/mL in H1703. IFNγ was treated at 100 ng/mL for 24 hours (left). Amivantamab reduced the expression of these genes under IFNγ-induced conditions (right).

Figure 6

Upregulation of EGFR and MET in HNSCC PDX (YHIM-3003) tumor induced increased expression of immune checkpoints regulators in the EGFR^HIGH/MET^HIGH subcluster (EM^HIGH). A, Volcano plot of top 50 genes in EGFR^HIGH/MET^HIGH against EGFR^LOW/MET^LOW tumor subcluster analyzed by log2 fold change (FC) against P-values. Red dots indicating transcripts with significantly increased fold changes including MET, PD-L1, and MET-regulated genes. B, Expression of STAT-4/PD-L1 (MET, STAT4, CD274), MET-regulated (BACE2, STK40, PRSS23, DPYD, CAV1, S100A4, PYGL), and MET-related immune checkpoints (HAVCR2, CD276) generally increased in the EGFR^HIGH/MET^HIGH tumor subcluster compared to the EGFR^LOW/MET^LOW subcluster. C, Expression of MET-related markers in different treatment groups.