AI-powered immune profiling from histopathology slides for chemo-radiotherapy outcome prediction in rectal cancer: a study using clinical trial and real-world cohorts

Background

The impact of the tumour-immune microenvironment on locally advanced rectal cancer (LARC) outcomes remains unclear. This study quantitatively assesses the synergistic influence of tumour-infiltrating lymphocytes (TILs), tumour-associated macrophages (TAMs), mitotic activity, and DNA mutations in predicting outcomes for LARC patients undergoing neoadjuvant chemo-radiotherapy (nCRT).

Methods

Three cohorts (ARISTOTLE-RC, UCLH-RC, TCGA-CRC) were stratified by densities of AI-quantified TILs, TAMs, and mitotic figures with cut-offs identified on a hold-out subset and integrated with DNA mutations to assess correlations with disease-free survival (DFS) and overall survival (OS). Immune cell dynamics pre- and post-CRT were also evaluated.

Findings

In ARISTOTLE-RC, TIL⁺ patients had significantly improved DFS (HR = 0.59, 95% CI: 0.39–0.90, p = 0.013) and OS (HR = 0.42, 95% CI: 0.24–0.73, p < 0.005), while TAM⁺ was associated with shorter DFS (HR = 1.65, 95% CI: 1.00–2.72, p = 0.045). Similar patterns were observed in UCLH-RC and TCGA-CRC. TIL⁺/KRAS⁻ patients had significantly improved DFS (HR = 0.41, 95% CI: 0.22–0.75, p < 0.005) and OS (HR = 0.28, 95% CI: 0.13–0.62, p < 0.005). In TP53-mutated patients, TAM⁺ group showed shorter DFS (HR = 1.46, 95% CI: 1.07–2.01, p = 0.0151), while among TP53-wild-type patients, no difference in DFS (HR = 1.00, 95% CI: 0.66–1.52, p = 0.9930) was observed between the two subgroups. Patients who transitioned after nCRT from TIL⁻ to TIL⁺ had improved DFS (HR = 0.70, 95% CI: 0.50–0.97, p = 0.028) and exhibited a significantly higher pre-treatment mitotic index (mean difference = 9.36, 95% CI: 1.87–16.85, p = 0.0385).

Interpretation

These findings suggest the potential utility of AI-driven immune profiling for clinical decision-making in LARC patients undergoing nCRT.

Funding

Cancer Research UK (RRNPSF-Jan21/100001, A18745, C7893/A2899), UK Research and Innovation (MR/T040785/1).