Researcher trying to take uncertainty out of cancer diagnosis
Cancer researcher Dr. Jason Sheltzer, of the Cold Spring Harbor Laboratory, hopes to take the uncertainty out of a cancer diagnosis. He and a colleague from Google have studied nearly 18,000 cancer cases, examining their medical histories and genetic data. They found that having a cancer-causing gene does not mean the cancer will turn deadly. Credit: Newsday/Alejandra Villa Loarca
A diagnosis of cancer tells a patient nothing more than the disease is present and doesn't reveal much about the long-term prognosis, uncertainty that can haunt patients sometimes for years, said a Cold Spring Harbor Laboratory researcher who believes he's on the cusp of erasing that fear.
Jason Sheltzer, a research fellow at the laboratory, has uncovered specific signposts — biomarkers — in tumors that can reveal a patient's long-term prognosis.
“We were looking at different types of cancer and changes in the letters of the genetic code. And we found these very strong prognostic patterns,” Sheltzer said.
The aim of his research is to take the uncertainty out of a cancer diagnosis.
He and his colleagues, including a team member from Google in California, analyzed genetic data from 17,879 tumors, along with the patients' medical records. All had been diagnosed with a variety of cancers.
Sheltzer discovered that even though a patient may have been diagnosed with a mutation linked to a specific malignancy, the cancer-related flaw was not predictive of who was most likely to survive for decades and who could wind up dying of the disease. He and his team examined prostate, breast, lung and other forms of cancer with known mutations.
Beyond the Cold Spring Harbor research, doctors long have relied on a variety of sophisticated tools to determine how fast a tumor is proliferating, for example, or how virulent a certain cancer will be, based on a range of biological factors. A gene-based test called Oncotype DX can even determine if a breast cancer patient will have a recurrence.
For more than two decades, pathologists have examined breast cancers' S-phase fraction, a measure that reveals the percentage of cells in a tumor that are in the process of replicating their DNA. The S-phase, which means synthesizing phase, is important because it occurs just before a cancer cell divides into two new cancer cells. This measure helps doctors understand how fast a tumor is growing.
In prostate cancer, doctors have relied on the Gleason score, two measures, each ranging from 1 to 5 that can determine the aggressiveness of a patient's tumor. A pathologist examines a biopsy's cells and analyzes whether they appear normal or more cancerous. An abundance of abnormal cells corresponds with a higher score and a poorer prognosis.
Sheltzer insists there is more information that can be gleaned from a tumor and the data not only can produce a prognosis, but provide the basis for better-targeted, precision therapy.
“There are hundreds of mutations that cause cancer,” he said, but these genetic miscues alone — with noteworthy exceptions — have not allowed doctors to predict a patient's prognosis.
Sheltzer was able to uncover identifiable biomarkers, which he calls “copy number alterations." This discovery allowed him to determine patients’ outcomes regardless of the type of cancer the patient had. The biomarkers provided telltale information, Sheltzer said, on how aggressive a patient’s cancer would be — or would not be.
He and his team collaborated with Joan C. Smith, a Google software engineer who moonlights in the arcane discipline of bioinformatics. She aided Sheltzer by studying data in the research. Bioinformatics is the science of analyzing complex biological information, particularly genetic data.
“I did all of the software engineering,” Smith said. “This project required crunching numbers in new and unique ways.”
Sheltzer said even he was surprised how the biomarkers — the copy number alteration molecules — harbored such significant prognostic power. The research, he added, allowed him to tap into a deep reservoir of data that tumors, once analyzed, can reveal about themselves.
In a healthy human cell, genes are distributed over 23 pairs of chromosomes. Cancer cells, however, usually gain or lose chromosomes in their helter-skelter pattern of growth.
Many cancers, instead of having two copies of a gene, will have three, four or even five copies, Sheltzer said, explaining what a copy number alteration is.
He hopes to further test the predictive capacity of copy number alterations in the ongoing collaboration between Cold Spring Harbor Laboratory and Northwell Health system’s cancer division.
