Cancer drug efficacy has remained a critical obstacle for researchers as it has one of the lowest probabilities of success compared to other diseases. One method to help improve this success rate is to create better tumor models on which to perform the drug testing. With growing interesting in microphysiological systems, scientists can create more advanced in vitro models of human organ systems as well as diseased states. These “organ-on-a-chip” platforms aim to improve drug response prediction for both efficacy and toxicity. One underappreciated characteristic of many disease states is that they are often at a lower oxygen tension that normal tissue (a condition known as hypoxia). Not only is this the case for cancer, but it is thought to be one of the main reasons why cancer treatment fails. Furthermore, cancer can experience two types of hypoxia:​ chronic (sustained low oxygen tension) or intermittent (varying cycles of hypoxia and normoxia). In both conditions, the vasculature determines the spatial and temporal dynamics of oxygen. If the tumor cells expand far enough away from the vasculature, this diffusion-limited oxygen condition will lead to chronic hypoxia. On the other hand, if the tumor cells encourage rapid vasculature growth that tends to be tortuous and leaky, then this will lead to intermittent hypoxia. Under both hypoxic conditions, tumor cells upregulate hypoxia inducible factor 1α (HIF-1α), a transcription factor which affects many downstream processes that encourages cell survival. In this dissertation, we investigate our ability to use a microfluidic platform to mimic some of these key features in the hypoxic tumor microenvironment and test if it can recapitulate some known biological responses.

First, we created a microfluidic device that can control both spatial and temporal gradients of oxygen via an oxygen scavenger line. As tumor angiogenesis is a major concern during tumor progression, we examined angiogenesis in the device. Vasculature was grown in the central chamber of the device, and stromal cells were grown in compartments on both sides of the vasculature. Lastly, the oxygen scavenger was flown in a channel adjacent to the left stromal chamber to create an oxygen gradient across the device. By controlling the flow the of oxygen scavenger, we could simulate both chronic and intermittent hypoxia. Our results demonstrate that stromal cells under hypoxia conditions encourage biased angiogenesis, which is in agreement with previous studies.

Second, using our microfluidic device platform, we investigated how knocking down HIF-1α affects the survival and progression of breast cancer cells (MDA-MB-231) under various oxygen gradients. By varying the scavenger concentration, we found a threshold for chronic hypoxia that the knockdown cells could no longer survive. Interestingly, by modulating the length of the hypoxic and normoxic cycles during intermittent hypoxia, we could dictate tumor cell survival. This result emphasizes the importance of understanding the temporal variations of oxygen, especially with cancer treatments that target HIF-1α.

Altogether, our results further our understanding of how to control spatial and temporal oxygen gradients for disease modeling. Hopefully with this understanding, future studies will be able to more effectively assess drug efficacy.