Characterising cancer-associated fibroblast heterogeneity in non-small cell lung cancer: a systematic review and meta-analysis

Abstract

Cancer-associated fibroblasts (CAFs) are a key component of the tumour microenvironment with evidence suggesting they represent a heterogeneous population. This study summarises the prognostic role of all proteins characterised in CAFs with immunohistochemistry in non-small cell lung cancer thus far. The functions of these proteins in cellular processes crucial to CAFs are also analysed. Five databases were searched to extract survival outcomes from published studies and statistical techniques, including a novel method, used to capture missing values from the literature. A total of 26 proteins were identified, 21 of which were combined into 7 common cellular processes key to CAFs. Quality assessments for sensitivity analyses were carried out for each study using the REMARK criteria whilst publication bias was assessed using funnel plots. Random effects models consistently identified the expression of podoplanin (Overall Survival (OS)/Disease-specific Survival (DSS), univariate analysis HR 2.25, 95% CIs 1.80-2.82) and α-SMA (OS/DSS, univariate analysis HR 2.11, 95% CIs 1.18-3.77) in CAFs as highly prognostic regardless of outcome measure or analysis method. Moreover, proteins involved in maintaining and generating the CAF phenotype (α-SMA, TGF-β and p-Smad2) proved highly significant after sensitivity analysis (HR 2.74, 95% CIs 1.74-4.33) supporting attempts at targeting this pathway for therapeutic benefit.

Figure 1

Flow chart describing steps carried out in selecting articles.

Figure 2

Analysis of individual markers. (A) Tree network showing number of studies for each marker per outcome group and analysis method. Figure generated using the vtree package in R (version 3.5.2). (B) Random-effect forest plots of individual markers from the OS/DSS outcome group and univariate analysis method.

Figure 3

Sub-group analysis of individual markers based on histological subtype. (A) Tree network showing number of studies for each marker per outcome group, analysis method and histological subtype. Figure generated using the vtree package in R (version 3.5.2). (B) Random-effect forest plots of individual markers from the OS/DSS outcome group and univariate analysis method with histological subtype indicated.

Figure 4

Ferris wheel plot summarising random-effect model HRs for each cellular process in CAFs. The height of each bar represents the HR for each process with the width of each bar indicating the % weight that each marker contributed to the random-effects model. The random-effect model HRs and 95% CIs are stated below each cellular process. Figure generated using Adobe Illustrator, 2020 (version 24.2). Icons representing each cellular process are from BioRender.com.