Cancer is a word no one wants to hear from the doctor. We’ve been fighting the disease for many decades now. The good news is that treatments have come a long way. In the past, cancer therapies are more like blunt instruments but at present, more and more therapies are becoming more targeted. Targeted therapies work only if we know what to target. To know that, one must understand the characteristics of the tumors. This brought me to the following question: how different or similar are breast tumors from ovarian tumors? They come from different parts of the body and arose from obviously different tissues. What molecular characteristics, such as genes that are mutated or overactive, are the same between breast and ovarian tumors?
Using the amazing amount of data amassed on tumors by TCGA (The Cancer Genome Atlas) community, I went ahead and made a map (network) using data from over 800 breast and ovarian tumors. The more similar the tumors are, the closer they will be in the map or network. I used both the genome information at the DNA level (what mutations occurred in the tumor) and at the RNA level (what genes are being expressed in the tumor).
Color scheme for top panels: red= indicated cancer; grey=not the indicated cancer. Color scheme for bottom panels: red=higher gene activity; blue= lower gene activity.
In the figure above, one can see that breast and ovarian tumors mostly do not overlap, with most of the breast tumors (top left panel, nodes colored red) residing on the right side of the tumor network and the ovarian tumors on the left side (top right panel, nodes colored red). This network is created from a fair number of tumor samples (870 tumors in all; about 500 of the tumors are from breast cancer patients, the rest are ovarian tumors). It is interesting to note that there are some breast tumors that fall in between the bulk of the breast and ovarian tumors. Upon further inspection, these tumors are what we call “Triple Negative Breast Cancer” (TNBC) tumors. These kinds of breast tumors are harder to treat than the other types of breast cancers. The bottom three panels of the figure above show lack of elevated gene levels of the hormonal receptors for estrogen (ESR1; the target of Tamoxifen) and progesterone (PGR) and lower levels of HER2/ERBB2 (the target of Herceptin).
Color scheme for top panel: red=mutation (1); dark blue=no mutation (0); color scale is the average between 0 and 1; basically the redder the nodes, the more tumors have a mutation in p53. Color scheme for bottom panels: red=higher gene activity; blue= lower gene activity.
So what are the common elements between the ovarian and the TNBC tumors? Mutations in p53 (TP53) is the one mutation that stood out. TP53 is what we call a tumor suppressor gene. It keeps an eye on our cell cycle and the fidelity of our genome. The top panel shows that both ovarian and TNBC tumors have mutations in TP53. TP53 is known to be mutated in many tumors; interestingly, you see that the non-TNBC tumors are mostly lacking this mutation but many of the TNBC and ovarian tumors do have this mutation. This gives us a good glimpse into how and when breast tumors are like ovarian tumors and when it is not. It is already known in the research community that cancer is a very heterogeneous disease. I am hopeful that as we look harder and harder into the differences and similarities of tumors that we are then able to tackle the heterogeneity better.
I also looked at what genes are active or inactive in the ovarian and the TNBC tumors compared to the other breast tumors. The level of gene activity of FOXA1 (bottom right panel) is lower in ovarian and TNBC tumors. FOXA1 binds to DNA and its expression is known to be highly correlated to the expression of the estrogen receptor gene (ESR1). It is not surprising to see its lower activity in TNBC tumors, which are not expressing much of the estrogen receptor gene.
On the hand, a gene called PSAT1 (bottom left panel) is especially active in the ovarian and TNBC tumors. PSAT1 is an interesting gene that has been shown to be active in colorectal cancer1. It is also an enzyme and therefore what we would call “targetable”, that is, it is possible to develop a compound against the protein product of this gene. It is not always true for any gene, so therefore, it is always fortunate to find something that is at least theoretically druggable. A pharmaceutical company would often start from there to develop a drug.
Food for thought: Fighting cancer has come a long way since the “declaration on the war with cancer” started over 40 years ago. Scientific communities, hospitals, commercial companies, patients and doctors are all coming together to painstakingly collect and analyze the data that have been and are being generated. I am excited about the medical revolution that is happening in my lifetime towards more personalized treatments, especially for cancer.
1. Mol Cancer. 2008 Jan 25;7:14. doi: 10.1186/1476-4598-7-14.