Is Risk-Stratifying Patients with Colorectal Cancer Using a Deep Learning-Based Prognostic Biomarker Cost-Effective?

Area of Science:

• Oncology
• Health Economics
• Artificial Intelligence in Medicine

Background:

• Accurate risk stratification for stage II and III colorectal cancer (CRC) is crucial for efficient adjuvant chemotherapy allocation.
• Limited health resources necessitate cost-effective methods to identify patients who will benefit most from treatment.

Purpose of the Study:

• To evaluate the cost-effectiveness of Histotyping, a deep learning-based prognostic method, for risk stratification in Norwegian colorectal cancer patients.
• To compare Histotyping against the standard of care (SOC) for adjuvant chemotherapy decisions in stage II and III CRC.

Main Methods:

• Development of two partitioned survival models for stage II and III CRC cohorts.
• Comparison of Histotyping-guided risk stratification with SOC (universal adjuvant chemotherapy).
• Measurement of health outcomes including quality-adjusted life years (QALYs) and life years (LYs) gained, with sensitivity and scenario analyses.

Main Results:

• Histotyping demonstrated a dominant strategy, being both less costly and more effective than SOC.
• For stage II CRC, Histotyping yielded a net monetary benefit of 270,934 NOK and a net health benefit of 0.99.
• For stage III CRC, Histotyping resulted in a net monetary benefit of 195,419 NOK and a net health benefit of 0.71.

Conclusions:

• Histotyping represents a likely cost-effective strategy for risk-stratifying colorectal cancer patients in Norway.
• The method aids in optimizing adjuvant chemotherapy administration, enhancing resource allocation and patient outcomes.