Impacting the Nature of Cancer Research

From machine learning predictions to personalized cell-based therapeutics—technologies propel cancer research

Dipanjan Chowdhury

Dipanjan Chowdhury, PhD
Associate Professor of Radiation Oncology
Harvard Medical School

Cancer research is being fundamentally changed in both experimental execution and data analysis.

The introduction of low-cost, high-throughput DNA sequencing has quickened the data-generating process. In a single day, sophisticated instruments run experiments transversing thousands of nucleic acids. As generating data is no longer a physically arduous process, the challenge now is finding the full meaning behind the data sets.

Dr. Chowdhury is creating a new blood test to rapidly detect early-stage ovarian cancer. This test detects small noncoding RNAs called microRNAs within the blood sample. Though microRNAs (defined as 22 base pairs) are smaller in size as compared to genes, there are still more than a thousand of these sequences located within the human cell. Working with microRNAs is no less complicated than a genome library—and data is rapidly generated with the same sequencing technology.

Chowdhury says, “Machine learning teaches the computer to recognize patterns. Instead of asking which individual genes contribute to ovarian cancer, we are looking for a signature combination of microRNAs.”

Chowdhury uses machine learning to leverage the sheer amount of data produced by sequencing blood samples from a patient with ovarian cancer. “We can feed the computer information from the thousands of microRNA molecules. We use statistics to understand the probability of microRNA levels found in patients with ovarian cancer versus patients without cancer cells or benign tumors. We don’t know how much of the combination of microRNAs contributes to the cancer itself. The algorithm gives the probability score of the risk of having ovarian cancer.”

Solving the data analysis bottleneck in cancer research may happen with machine learning, according to Chowdhury. He says, “I do believe the technology can be used in other cancers to produce similar success.”

In addition to machine learning, life science researchers are experimenting with ambitious therapeutic projects, especially in the development of personalized medicine.

Immunotherapeutics like chimeric antigen receptor (CAR) T cells genetically modify the patient’s own cells to target and destroy cancer cells found in leukemia. Gene editing is the key technology to generate CAR T cells for use in immunotherapy.

CAR T cell therapy has shown promise in treating leukemia, but many factors remain that hinder its application to other types of cancer. Challenges include the lack of unique antigens on solid tumor cells and the difficulty that
T cells have in infiltrating the solid tumor itself.

However, these problems may be addressed through improved antigen detection, changing gene editing strategies, and the use of more sophisticated technologies. Multiple academic and pharmaceutical laboratories are now working on improving the technology behind CAR T cell therapy.

These technologies and therapeutics are in their infancy, but the potential of applying machine learning and personalized medicine to cancer treatment is promising.

Subscribe to the Life in the Lab blog to keep current on the latest breakthroughs in CAR T research.