NK cell infiltration is associated with improved overall survival in solid cancers: A systematic review and meta-analysis

Abstract

The immune landscape of a tumor is highly connected to patient prognosis and response to treatment, but little is known about how natural killer (NK) cells predict overall survival (OS) among patients with solid tumors. We present the first meta-analysis on NK cell infiltration into solid tumors as a prognostic indicator for OS, considering cancer types independently, and together. Samples were collected from 1973 to 2016 with results published between 1989 and 2020. From 53 studies, we found that NK cell infiltration corresponds with decreased risk of death (HR=0.34, 95% CI: 0.26-0.46; p<0.0001). Among studies that investigated the prognostic potential of NK cells in specific regions of the tumor, intraepithelial infiltration was better predictive of OS than NK infiltration in the tumor-adjacent stroma. Generally, NK cell infiltration is lower in advanced-stage and lower-grade tumors; nevertheless, it remains prognostically beneficial. This meta-analysis highlights an important prognostic role of NK cells in solid tumors, but exposes that few studies have considered the contributions of NK cells. Toward NK cell-based immunotherapies, it will be important to understand the conditions under which NK cells can be effective agents of tumor control.

Keywords: Immuno-oncology; Natural killer cells; Solid tumor; Tumor infiltration; Tumor microenvironment.

Fig. 1

Flow-diagram outlining the process of study selection.

Fig. 2

Studies evaluating the associations between NK cell infiltration and overall survival. (A) Bar graph representing the distribution of conclusions from studies assessed in this review. From the 53 studies reviewed, the majority of studies (32 (59.3%)) reported significantly improved OS, 21 (38.9%) reported no significant impact on survival and one (1.9%) reported significantly poorer OS. (B) When p-values were provided (y-axis, line at 0.05) we noted them on this visualization scatter graph organized by tumor type (x-axis). Those that did not provide p-value are included above the graph. Dot size indicates the number of patients in each study (or arm of study when applicable).

Fig. 3

NK cell infiltration is associated with a decreased risk of dying in patients with solid tumors. (A) A random effects model meta-analysis was conducted on the 30 studies revealing a decreased risk of death in patients with greater NK cell infiltration (HR=0.34, 95% CI: 0.26–0.46; p<0.0001). Forest plots demonstrate pooled meta-analysis results by solid tumor type and pooled meta-analysis results from all solid tumor types. (B) All studies which reported HR values were visualized by Forest plot, grouped by tumor type (dots indicate the size of patient population studied).

Fig. 4

NK cell infiltration is highest in early stage and high grade tumors. Studies reporting NK cell infiltration at different (A, B) stages and (C, D) grades were included. The blue triangles indicate increasing or decreasing NK cell infiltration as stages and grades advance. Dot size indicates the number of patients in each study. (A) Studies that demonstrate increased NK cell infiltration corresponding with higher stage (n = 3). (B) Studies that demonstrate increased NK cell infiltration corresponding with lower stage (n = 9). (C) Studies that demonstrate increased NK cell infiltration at higher grades (n = 5). (D) Studies that demonstrate increased NK cell infiltration at lower grades (n = 2).

Fig. 5

The associations concluded for the impact of NK cell infiltration and may be influenced by IHC marker. (A) Forest plot visualizing the reported HR's of studies organized by marker used; NKp46 (orange), CD56 (dark blue), CD56 & CD57 (yellow) or CD57 (light blue). A random-effects model meta-analyses was conducted on studies using each marker. (B) This Forrest plot demonstrates the difference in the pooled risk of dying was larger in studies staining by CD56 (n = 16, HR:0.27, 0.18–0.41, p = 0.0001) and CD57 (n = 12, HR:0.38, 0.23–0.63, p = 0.0014) than NKp46 (n = 4, HR:0.58, 0.40–0.85, p = 0.020). (C) Bar graph demonstrating difference in proportion of studies finding associations between NK cell infiltration and survival when separated by marker used.