The rapid and widespread dissemination of fake news across digital platforms poses a serious threat to public trust in online media. The fake news often spreads rapidly because its complex and context-dependent nature, which enables it to bypass the traditional detection systems that lack deeper contextual understanding of information shared on news websites and social media. Existing detection approaches, particularly traditional content-based models, struggle to capture the interplay between textual content, social propagation patterns, and critical contextual or emotional cues. Our research addresses this critical gap by introducing a novel Context-Aware Sentiment Driven (gCASD) Fake News Detection framework that integrates Graph learning with Sentiment-Attuned Transformers to capture the deeper context and temporal feature for better detection. Extensive experiments show that the framework achieves strong performance, reaching validation accuracy of 94.13% on Politifact dataset and 88.54% and GossipCop dataset. These results consistently outperform other established methods, offering a more robust detection against fake news.