On Cancer: More-Accurate Colorectal Cancer Staging Helps Doctors Personalize Treatment

Memorial Sloan-Kettering researchers have pioneered prediction tools to assess disease risk and survival rates following surgery to remove colorectal cancer. These mathematical models, which are available online, are proving to be more accurate than traditional colorectal cancer staging methods.

Patients and physicians can use these prediction tools, known as nomograms, to help make treatment decisions, and researchers can use them to help plan clinical trials. A new nomogram, now available on Memorial Sloan-Kettering’s website, predicts a patient’s chance of survival – given in a percentage – five years after surgical removal of all cancerous tissue.

“As our therapeutic options have expanded, we need more-accurate predictions of disease survival and recurrence in order to make better decisions about treatment and long-term follow-up,” says Memorial Sloan-Kettering surgical oncologist Martin R. Weiser, who has led the effort to build colorectal cancer prediction tools. “Creating these new models enables us to address the distinct needs of every patient.”

Limitations of Traditional Colorectal Cancer Staging

The traditional approach to evaluating colorectal cancer risk, developed by the American Joint Committee on Cancer (AJCC), stages colorectal cancer based on minimal information about the tumor — specifically, three anatomical details regarding how much the cancer has grown and spread. The staging system is called TNM because it uses information about the tumor, lymph nodes, and metastasis (if the cancer has spread).

The TNM model categorizes cancer into one of four main stages, each of which is associated with a range of likely survival five years after surgery. In 2009, the AJCC divided TNM’s four stages into substages by incorporating more tumor characteristics.

The TNM model has shortcomings, Dr. Weiser explains. The outlook for different patients within the same overall stage can vary significantly. And although the addition of the substages improved the accuracy of prognosis within each stage, the clear order of the stages was lost — for example, some TNM stage III patients now have better survival rates than certain stage II patients.

“The AJCC’s increased number of categories has become confusing and often counterintuitive,” Dr. Weiser says. “We decided what was needed are predictive models that do away with all the groupings and instead produce for each patient an absolute number — a percentage showing the likelihood of remaining disease-free or surviving a certain number of years.”

More Data and Better Results

Memorial Sloan-Kettering’s nomograms build on the TNM system, supplementing it with routinely available data such as the patient’s sex and age, how much the tumor resembles its tissue of origin – known as tumor differentiation – and whether tumor cells were found in or around surrounding lymph nodes, blood vessels, and nerves. 

“The additional information gives our models a great deal of predictive power,” Dr. Weiser says. “And people easily grasp percentages, which makes it easier for patients and physicians to discuss treatment options.”

An earlier nomogram, which is also available online, predicts the probability of being disease-free five to ten years after surgery. The newer nomogram predicts a person’s chances of survival five years after surgery.

The researchers validated the newer nomogram’s accuracy by applying it to more than 125,000 colorectal cancer cases recorded in a government database. In December 2011, the researchers reported in the Journal of Clinical Oncology that the nomogram’s predictive power surpassed the TNM model when applied to the same database.

“I think this nomogram will more likely be seen as a valuable add-on that we offer here, rather than a complete replacement for traditional staging,” Dr. Weiser says. He explains that the nomogram will probably have the greatest impact on whether to use chemotherapy following surgery.

But the nomogram also could enable researchers to conduct clinical trials with fewer people, since more precisely defined risk groups make it possible to test a treatment’s effectiveness in a smaller population.

Dr. Weiser says the next step will be to incorporate molecular and genetic information into the prediction tools — much of it uncovered through research at Memorial Sloan-Kettering. “The biology of this cancer is much more complex than we imagined, so adding these new variables will be essential to refining these tools to make them as accurate as possible,” he explains.