Convergent CDK4/6 and PI3K/mTOR pathway hyperactivation defines a targetable axis in osteosarcoma across RB1-proficient and RB1-deficient contexts

Osteosarcoma (OS) in pediatric, adolescent, and young adult (AYA) patients is an aggressive malignancy with limited therapeutic progress. About 40% of OS patients develop metastases over time with 15-20% of OS patients already exhibiting metastases at initial diagnosis. Thus, there is a critical need to develop better therapeutics. Genomic analyses from our institution and others highlight recurrent alterations in CDKN2A and CDK4/6, suggesting that CDK4/6 inhibition (CDK4/6i) is a rational therapeutic vulnerability. Although RB1 proficiency (RB1+) is considered essential for CDK4/6i response, over 70% of OS tumors are RB1-deficient (RB1-), raising questions about the utility of CDK4/6i in these patients. Emerging evidence in other solid tumors and in our preliminary OS studies, suggests that CDK4/6i may retain antitumor activity in RB1- contexts. Compounding this complexity, pharmacologic CDK4/6 blockade induces compensatory PI3K/mTOR signaling, which can restore cyclin D-CDK4/6 activity and drive resistance. We hypothesized that co-targeting PI3K/mTOR would enhance CDK4/6i efficacy irrespective of RB1 status by suppressing adaptive signaling. We evaluated palbociclib, voxtalisib, and the combination across RB1+ and RB1- OS models, using cell lines, patient derived xenografts (PDXs), and an experimental metastasis model. In vitro drug interactions (Chou-Talalay, Bliss) and mechanistic assays (cell cycle, senescence, autophagy) were complemented by in vivo tumor growth kinetics and pharmacodynamic profiling through histopathology, kinome, and proteomic analyses. In RB1+ OS cells in vitro, palbociclib induced G1 arrest and senescence, accompanied by increased PI3K/AKT phosphorylation consistent with adaptive feedback. Voxtalisib suppressed this response, reinforced autophagic signaling, and maintained pathway inhibition. In vivo, combination therapy was well tolerated and produced significant tumor growth suppression in treatment-naïve and metastatic PDXs. In the RB1+ lung-colonization model, CDK4/6i alone reduced metastatic burden, with combination therapy achieving comparable control. To interrogate RB1 as a biomarker, CRISPR-engineered human and mouse RB1- OS clones (MG63.3, K7M2) were generated and characterized. As expected, voxtalisib response was RB1-independent, and palbociclib sensitivity was reduced in RB1- cells. However, RB1- cells still exhibited growth inhibition at higher palbociclib concentrations. Notably, combination therapy of palbociclib+voxtalisib produced additive-to-synergistic growth inhibition regardless of RB1 status. In vivo validation studies are ongoing. These findings identify convergent CDK4/6-PI3K/mTOR hyperactivation as a targetable axis in OS and support further evaluation of CDK4/6i-based strategies, including in RB1-deficient disease.