CT signature predicts survival, chemotherapy benefit in gastric cancer

Last Updated: 2020-02-03

By Will Boggs MD

NEW YORK (Reuters Health) - A deep-learning signature based on preoperative CT images predicts survival and the benefit of adjuvant chemotherapy in patients with gastric cancer, researchers report.

"Deep-learning analysis of diagnostic CT scans can reveal subtle image features that reflect underlying disease biology related to tumor progression and treatment response," Dr. Ruijiang Li of Stanford University School of Medicine, in Stanford, California, told Reuters Health by email.

Adjuvant chemotherapy is recommended for certain patients with stage-II gastric cancer with high-risk features and for stage-III patients, but standard clinicopathological risk factors do not consistently predict individual outcomes.

Dr. Li and colleagues used data from 1,615 patients with resected gastric cancer to develop and validate a deep-learning-based CT signature (DeLIS) to predicts disease-free survival (DFS) and overall survival (OS). They evaluated whether the signature could identify patients most likely to benefit from postoperative adjuvant chemotherapy.

At an optimal cutoff score, patients with a low DeLIS had an 85% lower risk of mortality and a 78% lower risk of progression (as measured by DFS), the team reports in the Annals of Surgery.

The five-year OS and DFS rates were 30.4% and 24.0%, respectively, for patients with a high DeLIS, compared with 66.1% and 61.2%, respectively, for patients with a low DeLIS.

Integration of DeLIS with four clinicopathological risk factors significantly improved the prediction of DFS and OS, compared with the clinicopathological risk factors alone.

In a propensity-score-matched analysis, adjuvant chemotherapy was associated with a significantly improved DFS for both stage-II and stage-III disease in the high-DeLIS group. Adjuvant chemotherapy did not affect DFS for either stage-II or stage-III disease in the low-DeLIS group, however.

In the intermediate-DeLIS group, patients with stage-II disease did not benefit from adjuvant chemotherapy, whereas patients with stage III did.

"The imaging signature is complementary to established clinicopathologic risk factors, such as disease stage," Dr. Li said. "Their combination improves prognostic prediction and could better select high-risk patients who will benefit from adjuvant chemotherapy."

"The imaging signature (derived from GE scanners) was validated in patients from two large medical institutions in Asia," he said. "It remains to be seen how it performs on western populations, or on other types of CT scanners."

"Its clinical use for therapy selection will require prospective validation in randomized controlled trials," Dr. Li added.

SOURCE: https://bit.ly/2GGdNru Annals of Surgery, online January 6, 2020.