P14/ARF-Positive Malignant Pleural Mesothelioma: A Phenotype With Distinct Immune Microenvironment

INTRODUCTION: The CDKN2A gene plays a central role in the pathogenesis of malignant pleural mesothelioma (MPM). The gene encodes for two tumor suppressor proteins, p16/INK4A and p14/ARF, frequently lost in MPM tumors. The exact role of p14/ARF in MPM and overall its correlation with the immune microenvironment is unknown. We aimed to determine whether there is a relationship between p14/ARF expression, tumor morphological features, and the inflammatory tumor microenvironment. METHODS: Diagnostic biopsies from 76 chemo-naive MPMs were evaluated. Pathological assessments of histotype, necrosis, inflammation, grading, and mitosis were performed. We evaluated p14/ARF, PD-L1 (tumor proportion score, TPS), and Ki-67 (percentage) by immunohistochemistry. Inflammatory cell components (CD3+, CD4+, CD8+ T lymphocytes; CD20+ B-lymphocytes; CD68+ and CD163+ macrophages) were quantified as percentages of positive cells, distinguishing between intratumoral and peritumoral areas. The expression of p14/ARF was associated with several clinical and pathological characteristics. A random forest-based machine-learning algorithm (Boruta) was implemented to identify which variables were associated with p14/ARF expression. RESULTS: p14/ARF was evaluated in 68 patients who had a sufficient number of tumor cells. Strong positivity was detected in 14 patients (21%) (11 epithelioid and 3 biphasic MPMs). At univariate analysis, p14/ARF-positive epithelioid mesotheliomas showed higher nuclear grade (G3) (p = 0.023) and higher PD-L1 expression (≥50%) (p = 0.042). The percentages of CD4 and CD163 in peritumoral areas were respectively higher and lower in p14/ARF positive tumors but did not reach statistical significance with our sample size (both p = 0.066). The Boruta algorithm confirmed the predictive value of PD-L1 percentage for p14/ARF expression in all histotypes. CONCLUSIONS: p14/ARF-positive epithelioid mesotheliomas may mark a more aggressive pathological phenotype (higher nuclear grade and PD-L1 expression). Considering the results regarding the tumor immune microenvironment, p14/ARF-negative tumors seem to have an immune microenvironment less sensitive to immune checkpoint inhibitors, being associated with low PD-L1 and CD4 expression, and high CD163 percentage. The association between p14/ARF-positive MPMs and PD-L1 expression suggests a possible interaction of the two pathways. Confirmation of our preliminary results could be important for patient selection and recruitment in future clinical trials with anticancer immunotherapy.