AI-identified histologic feature may offer ‘prognostic value,’ inform care in colon cancer
Click Here to Manage Email Alerts
Key takeaways:
- Machine learning-derived histologic feature may provide prognostic value when learned and scored by pathologists.
- While this represents a "milestone for AI," research is needed to validate reproducible scoring.
A prognostic study validated the use of an artificial intelligence-derived tumor adipose feature, which may aid risk stratification of colon cancer and provide prognostic value when integrated into practice, according to study results.
“Prognostic markers are of significant clinical interest in colorectal cancer, as some patients with stage II disease may benefit from adjuvant chemotherapy, and, for patients with stage III disease, improved prognostic information can inform treatment regimen and duration,” Vincenzo L’Imperio, MD, assistant professor of pathology at the University of Milano-Bicocca, and colleagues wrote in JAMA Network Open. “In this setting, the use of digital pathology tools has recently demonstrated the capability to provide prognostic information about colon cancer with the use of routine histopathologic slides.
Image: Adobe Stock
“This led to the identification of the tumor adipose feature (TAF), moderately to poorly differentiated tumor cells in close proximity to adipocytes, as a machine learning-derived feature that demonstrated promising, independent prognostic value in stage II and III colorectal cancer cases.”
L’Imperio and colleagues used data from 258 colon adenocarcinoma histopathologic cases (53% men; median age, 67 years; stage II, n = 119; stage III, n = 139) to investigate whether pathologist scoring of histopathologic features previously identified with machine learning correlated with survival.
Two pathologists, who were blinded to the patient outcomes, identified TAF in 47% of cases, with multifocal involvement in 12% and widespread involvement in 24%. Pathologist agreement was 72% across all TAF scores and 90% for widespread TAF compared with other classification.
Researchers reported “significant prognostic value” of pathologist-identified TAF using a binary threshold for overall survival (HR = 1.55; 95% CI, 1.07-2.25) but not for CRC disease-specific survival (HR = 1.86; 95% CI, 0.95-3.62). However, there was a quantity-dependent association with widespread TAF and overall survival (HR = 1.87; 95% CI, 1.23-2.85) as well as disease-specific survival (HR = 2.29; 95% CI, 1.09-4.7).
In multivariable analysis, age (HR = 1.07; 95% CI, 1.05-1.09), stage (HR = 1.6; 95% CI, 1.03-2.51) and widespread TAF (HR = 1.79; 95% CI, 1.14-2.81) remained independently prognostic for overall survival, whereas stage (HR = 3.57; 95% CI, 1.39-9.18) and widespread TAF (HR = 2.19; 95% CI, 1.01-4.75) remined independently prognostic for disease-specific survival.
“This prognostic study represents a milestone for AI in pathology and medicine, demonstrating both the feasibility and prognostic potential for pathologist-based integration of a feature identified via machine learning,” L’Imperio and colleagues concluded. “After the demonstration of generalizable prognostic value and consistent scoring strategies across pathologists, AI-derived prognostic features can potentially be used along with well-established features in prospective cases to enable further validation and clinical integration.”
Collapse
Read more about
Click Here to Manage Email Alerts