Chris Bakal, PhD, leads the Dynamical Cell Systems team at the Institute of Cancer Research in London, England. He and his team recently created a “map” linking the shape of breast cancer cells to how genes turn on and off, and matched it to real disease outcomes, which could help clinicians select treatments. Details of this research were published recently in Genome Research.1

In this question-and-answer session, Cancer Therapy Advisor asked Dr Bakal about this new approach to analyzing cell shape, its potential to affect treatment for breast cancer, and the future directions of using big data to enrich cancer research.

Continue Reading

Cancer Therapy Advisor (CTA): Looking at the appearance of cancer cells is not new. Until now, what information could pathologists possibly gather from looking at a cancer cell’s shape?

Dr Bakal: The answer is twofold. In the clinic, pathology is used as a diagnostic tool to predict what is going to happen to the patient. Pathologists look at different features or aspects of cell shape, which is reflected in tissue morphology, and can tell whether a cancer is serious, if prognosis is poor, whether it is going to be well-treated, easily resectable, or easy to deal with.

In the lab, researchers can get a sense of cellular behaviors they are studying. They can infer whether a cell might be more proliferative, more metastatic or invasive, or drug resistant. They get a sense of what a cell, or population of tumor cells is going to do in the future.

A lot of pathology has a very “human” aspect to it. Pathology is a human looking down a microscope and making judgments based on training and experience. Many pathologists with lots of experience look at images and know how the cell shape or tissue morphology is likely to be predictive of cancer outcomes.

CTA: By using “big data” you were able to analyze millions of images at once. What did this analysis reveal?

Dr Bakal: Our analysis revealed 3 things. First, and perhaps a little bit of a surprise, was the extent to which transcription — the way a gene is turned on and off — correlates with cell shape. It has been appreciated, both in clinical and fundamental research, that there is a link between shape and transcription, but our study suggested that the link between cell shape and transcription is broader and wider then we appreciated before we started this study.

The second major point of our analysis was that not only was there a link between cell shape and transcription, but the link works both ways and in complicated fashions. Not only does transcription regulate shape, but there is also a large effect of cell shape on transcription. The shape of a cell dictates transcriptional activity.

We also found that shape and transcription are linked to the activity of a transcription factor, NF-kappaB. This was interesting because NF-kappaB regulates inflammation. Whenever there is an immune response or a wound in the cell, NF-kappaB is turned on as a way to repair the wound, deal with the stress response, and recruit immune cells. It is a pro-inflammatory factor. The fact that it is also linked to cell shape was very interesting. In breast cancer, we know that NF-kappaB is a driver of cancer. In many patients, we know there are high levels of NF-kappab activity, but it is very rarely mutated in solid tumors. This suggests that changing the shape of a cell can turn on inflammation to drive cancer in the absence of any mutation.

Finally, the link we found between shape, transcription and NF-kappaB was very relevant to cancer prognosis. The network we described, the genes we identified are very much predictors of outcomes in breast cancer patients. We could have found a link between shape and transcription, but what we saw using the clinical data is that this link is very important to the patient. You can, in fact, predict cancer outcomes based on much of the actions of this network alone. So this relationship between shape and transcription is very important to disease progression — more so than was known before.