Recent advances in Generative Adversarial Networks (GANs) have led to many new variants and uses of GANs. The latest advancements have allowed researchers and practitioners to apply this technique to tumor-related problems with limited data. One of the trends in this problem domain is to develop different variants of GANs suited explicitly to particular problems. The variants of GANs are numerous but share a common characteristic of expanding the dataset by creating synthetic data from the original dataset. This paper aims to develop a research agenda through a systematic literature review that investigates practitioners' and researchers' emerging issues and current works on the topic. Emerging implementation trends and limitations of GANs in tumor-related problems are explored.