Multi-omics Analysis Reveals Immune Features Associated with Immunotherapy Benefit in Patients with Squamous Cell Lung Cancer from Phase III Lung-MAP S1400I Trial

Abstract

Purpose: Identifying molecular and immune features to guide immune checkpoint inhibitor (ICI)-based regimens remains an unmet clinical need.

Experimental design: Tissue and longitudinal blood specimens from phase III trial S1400I in patients with metastatic squamous non-small cell carcinoma (SqNSCLC) treated with nivolumab monotherapy (nivo) or nivolumab plus ipilimumab (nivo+ipi) were subjected to multi-omics analyses including multiplex immunofluorescence (mIF), nCounter PanCancer Immune Profiling Panel, whole-exome sequencing, and Olink.

Results: Higher immune scores from immune gene expression profiling or immune cell infiltration by mIF were associated with response to ICIs and improved survival, except regulatory T cells, which were associated with worse overall survival (OS) for patients receiving nivo+ipi. Immune cell density and closer proximity of CD8+GZB+ T cells to malignant cells were associated with superior progression-free survival and OS. The cold immune landscape of NSCLC was associated with a higher level of chromosomal copy-number variation (CNV) burden. Patients with LRP1B-mutant tumors had a shorter survival than patients with LRP1B-wild-type tumors. Olink assays revealed soluble proteins such as LAMP3 increased in responders while IL6 and CXCL13 increased in nonresponders. Upregulation of serum CXCL13, MMP12, CSF-1, and IL8 were associated with worse survival before radiologic progression.

Conclusions: The frequency, distribution, and clustering of immune cells relative to malignant ones can impact ICI efficacy in patients with SqNSCLC. High CNV burden may contribute to the cold immune microenvironment. Soluble inflammation/immune-related proteins in the blood have the potential to monitor therapeutic benefit from ICI treatment in patients with SqNSCLC.

Figure 1.

Kaplan–Meier survival curves of cellular densities and immune signatures. In the nivo arm, Kaplan–Meier survival curves show high cellular densities (>the median value used as cutoff) of memory T cells (CD3+CD45RO+; A) in the total compartment and CD45RO+ Tregs (CD3+CD45RO+FOXP3+; B) in the stroma compartment were associated with better PFS. C, CD45RO+ Tregs (CD3+CD45RO+FOXP3+) in the total compartment were associated with better OS. Representative multispectral images show low and high cell phenotype densities for A–C. In the nivo+ipi arm, the Kaplan–Meier survival curves show that high cellular densities of PD-1+ T cells (CD3+PD-1+; D) in the total compartment and GZB+ CTLs (CD3+CD8+GZB+; E) in the total compartment were associated with better PFS. Conversely, Tregs (CD3+CD8−FOXP3+; F) in the total compartment were associated with poor OS. Representative multispectral images show low and high cell phenotype densities for D–F. Cell scoring derived from gene expression profiling using nSolver shows higher scores for CD45+ immune cells (G), CD8+ T cells (H), and neutrophils (I) in responders compared with nonresponders in the nivo+ipi arm.

Figure 2.

Immune infiltration in exceptional responders and early progression across the arms. A, The upper level of the event chart shows the exceptional responders, and the lower level shows the early progression/death group. The solid red circles represent deaths in the OS analysis, the open red circles indicate OS-censored patients, the solid blue triangles indicate progression in the PFS analysis, the open blue open triangles indicate PFS-censored patients, and the violet X indicates the time to the first response. Representative multispectral images of panels 1 (B) and 2 (C) show high levels of inflammatory cells in a sample from an exceptional responder patient. D, Box plot shows GZB+ CTLs (CD3+CD8+GZB+) in patients with exceptional response compared with patients with early progression/death. Representative multispectral images of panels 1 (E) and 2 (F) show reduced immune infiltration in a progression/death patient sample. G, The spatial organization of immune and malignant cell phenotypes for the two mIF panels is shown with an example each from exceptional responders and early progressors. The colors for the different subpopulations are indicated under the panel phenotype legend (on the left). H, For the above images, segregation of different cell phenotypes based on their spatially varying probabilities is shown as a contour plot. The colors of different neighborhoods are same as the panel phenotypes (above). I, For the above images, Euclidean distance–based clusters of cells (10 or more) within 20 μm are identified. The clusters are represented by numbers and distinct colors. J, The relative percentage composition of cell types within each cluster (above) is indicated in the heat map. The corresponding cluster colors are indicated below the heat map for reference. The color scale representing percentage composition (0–100) is shown on the left.

Figure 3.

Kaplan–Meier survival curves of nearest neighbor distance from both arms. A, Upper and lower image showing proximity map overlay, where cyan dots represent malignant cells (CK+) and red dots represent T cells (CD3+). White lines display distances from all malignant cells (CK+) to neighboring T cells (CD3+). Kaplan–Meier survival curves show that distances (≤the median value used as cutoff) from malignant cells (CK+) to CTLs (CD3+CD8+; B) and GZB+ CTLs (CD3+CD8+GZB+; C), and malignant cells expressing PD-L1 (CK+PD-L1) to CTLs (CD3+CD8+; D) were associated with better PFS when combining both treatment arms. E, Kaplan–Meier OS curve for distances from malignant (CK+) to GZB+ CTLs (CD3+CD8+GZB+) in both arms. In the nivo+ipi arm, Kaplan–Meier survival curves show that close distances (≤the median value used as cutoff) from malignant cells (CK+) to CTLs (CD3+CD8+; F) and GZB+ CTLs (CD3+CD8+GZB+; G), and PD-L1+ malignant cells (CK+PD-L1+) to CTLs (CD3+CD8+; H) were associated with better PFS. I, Close distances from malignant cells (CK+) to GZB+ CTLs (CD3+CD8+GZB+) was associated with OS.

Figure 4.

Olink serum soluble analyte assessment. A, Heat map of dynamic changes in protein expression. The x-axis shows the protein names, while the y-axis shows the comparisons between timepoints and progression. The color represents the logFC. Green represents increase from baseline while pink represents decrease. The size of the circle indicates the statistical significance expressed as –log₁₀(FDR). B, Boxplots and median rend lines showing the expression over time by cohort for CXCL9 and CXCL13. C, Heat map for differential protein expression between responders and nonresponders. The x-axis shows the protein names, while the y-axis shows each timepoint. The symbol in the heat map represents the statistical significance: circles for FDR < 0.05 or adjusted P values, squares for P < 0.05, and triangles for nonsignificant or P > 0.05. The color represents the change relative to upregulation in responders (blue) or nonresponders (red). D, Boxplots corresponding to significant markers in C over time, stratified by treatment arm for the indicated proteins. Comparisons for individual baseline, cycle 2, and cycle 4 timepoints are shown for P < 0.05 and FDR < 0.05 with (*) and (**), respectively. E, Heat map showing the concordance in directionally of differentially expressed proteins significant between exceptional responders and all responders. The direction of the protein changes was identical between both groups of responders, but only CXCL13 and CCL23 reached statistical significance (FDR, darker colors) for exceptional responders due the decreased numbers. Nominal significance is shown as transparent colors, indicating proteins with P < 0.05. F, Volcano plot showing the proteins significantly associated with OS when jointly modeling cytokine expression over time. The proteins labeled in blue are associated with increased HR or decreased survival. Kaplan–Meier OS curves for CSF-1, IL8, CXCL13, and MMP12 stratified on the basis of their expression from the average expression (higher values from the mean as blue, lower values from the mean as red).