Prior to my graduate research, I had been under the impression that most breast cancer patients would receive similar treatments in the form of chemotherapy and depending on the severity and spread of the disease; surgery.
However, breast cancer is a heterogeneous disease, at the cellular and molecular level we have been able to differentiate between breast cancer subtypes and associated these subtypes with specific characteristics. As a result, a patient suffering from a triple-negative subtype may not respond to hormonal therapy as a HER-2 type subtype. Even patients classified as a triple-negative subtype (in which case these patients can only receive chemotherapy treatment) may not respond positively to the treatment administered. In reality, we find that some patients respond to treatment and some don’t and currently there is no true assay or metric that can identify those who would benefit from those that would not.
But why is it necessary to find out which breast cancer patient can receive chemotherapy from those who cannot? What impact does this have with therapy? With our inability to confidently determine which patient will benefit from a particular therapy and by following the “one therapy to all” process, we increase the likelihood of side effects and eventual resistance to therapy. Similar to the concept of pests and weeds developing resistance to pesticides and passing down the resistant gene to their offspring, there is evidence that supports a similar trend in cancer cells, in fact, this resistance may develop from a number of different molecular activities (Housman G, et al., 2014).
In addition, we continue to develop novel treatments without a clear idea of who will benefit, as such the concept of personalized medicine and being able to customize treatment for patients is an imperative course of action.
The introduction and development of various assays that rely on specific genes to inform clinicians of breast cancer subtypes linking to appropriate therapy are currently on the rise. Projects such as the TAILOR-X and MINDACT are molecular profiling tests used to potentially determine appropriate therapy.
As a side project, I used a publicly available database of the breast cancer patients molecular profile (The Wisconsin Breast Cancer database) and organized the data through Python with the aim of identifying whether molecular characteristics of the breast cancer can determine malignancy. By identifying a trend/pattern we may be able to train systems/machines (within a supervised learning approach) to produce reliable outputs with every input (i.e., patient breast cancer subtype information). Next blog post will feature my initial analysis using Python.