Natural Killer Cell Infiltration in Prostate Cancers Predict Improved Patient Outcomes

Abstract

Background: Natural killer (NK) cells are non-antigen specific innate immune cells that can be redirected to targets of interest using multiple strategies, although none are currently FDA-approved. We sought to evaluate NK cell infiltration into tumors to develop an improved understanding of which histologies may be most amenable to NK cell-based therapies currently in the developmental pipeline.

Methods: DNA (targeted/whole-exome) and RNA (whole-transcriptome) sequencing was performed from tumors from 45 cancer types (N = 90,916 for all cancers and N = 3365 for prostate cancer) submitted to Caris Life Sciences. NK cell fractions and immune deconvolution were inferred from RNA-seq data using quanTIseq. Real-world overall survival (OS) and treatment status was determined and Kaplan-Meier estimates were calculated. Statistical significance was determined using X² and Mann-Whitney U tests, with corrections for multiple comparisons where appropriate.

Results: In both a pan-tumor and prostate cancer (PCa) -specific setting, we demonstrated that NK cells represent a substantial proportion of the total cellular infiltrate (median range 2-9% for all tumors). Higher NK cell infiltration was associated with improved OS in 28 of 45 cancer types, including (PCa). NK cell infiltration was negatively correlated with common driver mutations and androgen receptor variants (AR-V7) in primary prostate biopsies, while positively correlated with negative immune regulators. Higher levels of NK cell infiltration were associated with patterns consistent with a compensatory anti-inflammatory response.

Conclusions: Using the largest available dataset to date, we demonstrated that NK cells infiltrate a broad range of tumors, including both primary and metastatic PCa. NK cell infiltration is associated with improved PCa patient outcomes. This study demonstrates that NK cells are capable of trafficking to both primary and metastatic PCa and are a viable option for immunotherapy approaches moving forward. Future development of strategies to enhance tumor-infiltrating NK cell-mediated cytolytic activity and activation while limiting inhibitory pathways will be key.

Fig. 1. Characterization of NK cells in tumors.

A NK cell fraction as a percentage of all cells calculated using quanTiseq for immune deconvolution using transcriptomic data. A total of 90,916 samples from 45 distinct tumor types were analyzed from a commercial database of real-world patient samples. Data shown in violin plots in which the white dot represents the median and the black box shows the ends of the first and third quartiles. p * < 0.05, ** < 0.01, *** < 0.001, **** < 0.0001. B Distribution of quanTIseq NK cell score across PCa tumors from the prostate and metastatic PCa tumors. Vertical dotted lines represent the median NK cell fraction for each subgroup. C Metastasis was then divided by anatomical sites, including liver and bone. Cohorts shown are stratified by the median NK cell fraction. Distribution of NK cell scores shown in violin plots in which the boundary of the violin represents the range of all data points.

Fig. 2. NK cell profiles association with OS.

A Association of overall survival (OS) in tumors with top 75th percentile NK cell infiltration (relative to all cancers) across 45 different tumor types. Dotted line represents the Hazard Ratio of 1.0. p * < 0.05, ** < 0.01, *** < 0.001, **** < 0.0001. B Progression-free survival (PFS) of patients with high (>= median) NK infiltration compared to low (<median) NK infiltration in 45 distinct tumor types. C OS of patients with tumors from the prostate and metastasis were partitioned into 4 quartiles as defined by the relative NK cell abundance. D OS for metastasis was further divided by common anatomical metastatic sites, including liver and bone. Cox proportional hazard ratios were calculated for each comparison group with significance determined as p values of <0.05 using log-rank statistics.

Fig. 3. Interaction of NK cell profiles with driver mutations in PCa.

Pathogenic mutation frequency of common PCa drivers (A), median WTS (TPM) for AR associated genes (B), and pathogenic mutation frequency for AR associated genes (C) across NK cell high and low prostate and metastatic samples. p * < 0.05, ** < 0.01, *** < 0.001, **** < 0.0001. Cox proportional hazard ratios were calculated for each comparison group with significance determined as p values of <0.05 using log-rank statistics.

Fig. 4. NK cell association with current ICI targets.

Median WTS (TPM) for immune regulatory genes, some of which are immunotherapy targets, across NK cell high and low (A) prostate and (B) metastatic samples. p * < 0.05, ** < 0.01, *** < 0.001, **** < 0.0001. Cox proportional hazard ratios were calculated for each comparison group with significance determined as p values of <0.05 using log-rank statistics.

Fig. 5. NK interaction with other Immune cells.

A Immune cell deconvolution using quanTiseq demonstrates differences in cell fraction of B cells, macrophage M1/ M2, dendritic cells, neutrophils, T cell regulatory, and NK cells, as well as (B) monocytes, and T cells CD4/ CD8 in prostate and metastatic PCa tumors when comparing Q1 and Q4 NK cell infiltration groups. *p-values (**** < 0.0001).