Abstract
Background: Growth/Differentiation Factor 15 (GDF15) has been implicated in cancer carcinogenesis, neoangiogenesis, and cancer cachexia. High circulating or tissue protein expression levels of GDF15 have been associated with worse clinical outcomes in a wide range of tumor types, correlating with platinum resistance and immune exclusion in preclinical models. GDF15 antagonist antibodies are in clinical development for both cachexia and augmentation of immunotherapies. Since GDF15 data are limited in BTC, we sought to characterize GDF15 expression patterns, its impact on the tumor immune microenvironment, and prognosis in a large real-world cohort.
Methods: We used Tempus Lens to identify patients (pts) with a primary diagnosis of BTC who had Tempus xT DNA and xR RNA testing from the Tempus real-world multi-modal database (N = 4,479). RNA-seq data were normalized to correct for assay/batch effects, quantified as transcripts per million (TPM) and reported as log2(TPM+1). Pts were classified as GDF15 high (GDF15-H, n=1120) and GDF15 low (GDF15-L, n=1120) expressors based on the top and bottom quartile of GDF15 mRNA expression. Immune cell proportions and cytolytic, cytotoxic, and interferon-γ immune scores were estimated from RNA expression. Chi-squared/Fisher’s exact or Wilcoxon rank sum tests were used to assess statistical significance. Real-world overall survival (rwOS) was defined as the time from sample collection to death or loss to follow up and assessed using Cox proportional hazards models.
Results: BTC GDF15-H pts had a higher median expression compared to GDF15-L expressors (7.09 vs 3.32). GDF15-H cases comprised a higher proportion of intrahepatic CCA primaries (61% vs. 32%) and lower proportion of gallbladder cancer (GBC) compared to GDF15-L BTC samples (15% vs. 35%). GDF15-H pts had a higher frequency stage I/II disease (17.7% vs. 8.4%) and lower frequency of stage IV disease compared to GDF15-L BTC (69% vs. 78%). GDF15-H tumors also had a higher frequency of concomitant actionable mutations eligible for FDA-approved therapies (39% vs. 17%, p<0.001) and had significantly lower cytolytic, cytotoxic, and interferon-γ immune scores (all p<0.001) compared to GDF15-L tumors. On multivariable regression analysis, controlling for stage, TP53 status, and presence of actionable mutation, there was no significant difference in OS outcomes between GDF15-H and GDF15-L BTC cohorts.
Conclusion: In this large, real-world analysis of BTC, tumor GDF15 mRNA levels did not significantly correlate with rwOS. BTC pts with GDF15-H mRNA expression had less immunologically active tumors compared to GDF15-L tumors. These findings diverge from previously published literature on GDF15 as a prognostic biomarker/therapeutic target when it is assessed based on local or circulating protein level. Given our findings and the fundamental complexities of GDF15, including extensive post-translational processing and source of production, future studies are needed to reconcile our understanding of the functional and prognostic role of GDF15 in BTC.
VIEW THE PUBLICATION
VIEW THE POSTER