Detecting Sarcoma at the Right Time

New data-based research points to optimum CT-scan interval

By Kardelen Koldas ’15

February 25, 2021

Doctors often say early diagnosis saves lives, especially when it comes to cancer. For patients diagnosed with sarcoma—rare tumors that often metastasize to the lungs—surveillance methods like chest computed tomography (CT) are crucial to catching the disease’s spread. But what’s the optimal interval for chest CT?

Until recently, the answer stemmed from expert opinion, which can be subjective. Now, thanks to a study coauthored by Charles A. Dana Professor of Statistics Liam O’Brien, the answers come from data.

Using statistical modeling, the research showed that the optimal interval can be estimated based on the tumor’s growth rate. This knowledge promises to help treatment as doctors can intervene at the correct time before nodules grow out of control.

In the study, published in the British Journal of Radiology, O’Brien and his coauthors from the University of Pennsylvania, medical student Ulysses Isidro and Assistant Professor of Radiology Ronnie Sebro, examined successive chest CT scans of 95 sarcoma patients with pulmonary metastases. The records, spanning across two decades, contained information on patients’ age, sex, tumor size, tumor grade and subtype, and chemotherapy the researchers used in their analysis.

Their goal was to estimate pulmonary metastasis growth rate and determine the ideal frequency for chest CT to detect these sarcomas at the right time. They also sought to determine whether the growth rate varied by patient’s sarcoma subtype (there are more than 50) and whether all the nodules of a patient enlarged at the same rate.

“We had longitudinal data for which we were able to do these things,” said O’Brien. “And it’s just never been done before.”

O’Brien explained that the nodule needs to be at least two millimeters (0.08 inches) in any one dimension for CT to pinpoint its location. And at eight millimeters (0.31 inches), doctors deem the nodule “actionable.” The data indicated that it takes roughly 252 days, almost eight and a half months, for an untreated pulmonary metastasis to grow from two to eight millimeters. “If your interval is six months, then it’s probably about right,” said O’Brien.

The United States National Cancer Center Network (NCCN) recommends imaging every six to 12 months at stage I, he noted. This interval shortens as the stage becomes more advanced, “so it looks like the intervals are adequate because they’re all six months or less.”

The researchers also found that pulmonary nodule volume grows exponentially, meaning the tumor takes a shorter time to enlarge from four to six millimeters than from two to four millimeters. This is why CT scans become more frequent as the disease progresses.

Overall, the data revealed that each month sarcoma pulmonary metastases grow at approximately 0.7 millimeters (0.03 inches). Small (two millimeters) pulmonary metastases, on the other hand, grow about one-third of a millimeter (0.01 inches). Although the patient’s age, size, grade, or subtype didn’t affect the growth rate, chemotherapy proved to slow it down.

The researchers also looked at the growth rates on an individual basis. They identified that from one patient to the other, and even within the same patient, pulmonary metastasis grows at varying rates. “Each person’s sarcoma is unique—even the same type of tumor in the same person may grow at a different rate,” said O’Brien. “And that really has to do with the genetics of the tumor.”

Consequently, the paper put forth a cautionary tale, even as it confirmed the guidelines of the NCCN.

“If you’re going to try to do a one size fits all interval for everybody, be aware that there’s a lot of variation and that you’re going to miss some people,” said O’Brien. “That’s the subtle but important warning.”

O’Brien emphasized the importance of employing statistics in medicine. Whether to determine CT frequency for sarcoma patients or analyze clinical trial data for Covid-19 vaccines, statisticians play a key role. “Statistics,” he said, “is really what you need to meet these goals.”