New method of gauging metastatic risk may help ovarian-cancer patients avoid unnecessary chest CTs

The National Comprehensive Cancer Network recommends chest and abdominopelvic imaging, primarily with CT, for follow-up of most women with ovarian cancer. That guideline may now be due for a revisit, as researchers in the U.S. and South Korea have collaborated to create and validate an evidence-based rule that, in their retrospective review of several hundred patient cases, accurately predicted which ones were not at significant risk for thoracoabdominal metastases.

“The easy-to-use prediction rule helps avoid unnecessary chest CTs in patients with ovarian cancer with high sensitivity and negative predictive value, and with minimal risk of missing thoracoabdominal metastases,” the authors report in a study published online Oct. 6 in the Journal of the American College of Radiology.

Atul B. Shinagare, MD, of Harvard, Chong Hyun Suh, MD, of the University of Ulsan, and colleagues searched EMRs at Asan Medical Center in Seoul (January 2012 through December 2013) and Dana-Farber Cancer Institute/Brigham and Women’s Hospital in Boston (calendar year 2013) to identify consecutive chest CTs performed within five years of initial surgery in patients with pathologically proven ovarian cancer.

They broke out the cases to create three separate study cohorts: cohort one, 316 CTs (in 150 patients) with high-grade serous epithelial ovarian cancer (HGSC); cohort two, 374 CTs (81 patients) with HGSC; and cohort three, 87 CTs (56 patients) with non-HGSC histologies.

The researchers had radiologist who was blinded to outcome of CT use a prediction rule drawing from previously available EMR information to score each CT as either high-risk (score of 1 or 2) or low-risk (score of 0) for thoracoabdominal metastases.

The authors describe how the scoring worked:

Preexisting abdominal disease and stage 4 disease at presentation each were associated with a score of 1. Therefore, chest CT where there was no prior preexisting abdominal disease and the patient did not have stage 4 disease at presentation would get a score of 0; CT where one of these characteristics was present would get a score of 1; and if both preexisting abdominal disease and stage 4 at presentation were present, CT was assigned a score of 2.

The radiologist’s rule-based categorizations were then compared with reviews by a blinded radiologists of the chest CTs in random order.

The research team’s key findings:

  • The evidence-based prediction rule identified 94 of 316 (30 percent), 170 of 374 (45 percent) and 53 of 87 (61 percent) CTs as “low-risk” for thoracic metastases (neither stage 4 at presentation nor preexisting abdominal disease) in three separate patient cohorts.
  • Overall, 317 of 777 (41 percent) CTs were “low-risk.”
  • In the three cohorts and overall, the prediction rule had 95 percent, 88 percent, 88 percent, and 91 percent sensitivity and 97 percent, 95 percent, 98 percent, and 96 percent NPV, respectively.
  • The false-negative rate was 3 of 94 (3 percent), 8 of 170 (5 percent), 1 of 53 (2 percent), and 12 of 317 (4 percent). However, each CT showed concurrent new abdominal disease. “Thus, no new thoracoabdominal metastases were ‘missed,’ because there were no thoracic metastases without preexisting or concurrent abdominal disease,” the authors note.

The authors conclude by stressing that their prediction rule is evidence-based, easy to use and had a good sensitivity and negative predictive value in assessing the risk of thoracic metastases.

The false-negative rate was low, they add, and in each case there was new concurrent abdominal disease—a finding that would, in practice, guide clinical management.

“This prediction rule would increase the yield of chest CT, leading to better utilization,” the authors write, adding that it could help reduce imaging costs and radiation burden while still addressing the clinical needs of women with ovarian cancer.

The rule, they write, “may easily be incorporated into the clinical workflow as a clinical decision support tool to assist the oncologists in making decisions regarding chest CT based on easily available information, and can potentially avoid up to 41 percent of chest CTs with minimal, if any, risk of missing thoracoabdominal metastases.”