KEY POINTS
- Investigators analyzed pretreatment biopsies from three locally advanced rectal cancer cohorts: 193 patients from S:CORT Grampian, 200 from GSE87211 and 52 evaluable patients from the Beatson West of Scotland Cancer Centre. The first two underwent transcriptomic microarray profiling, while the third provided multiplex immunofluorescence and histologic validation.
- Approximately 95% of patients received long-course chemoradiotherapy. Regimens included 50 Gy/25 fractions, 50.4 Gy/28 fractions, or in a minority 25 Gy/5 fractions, with fluoropyrimidine-based chemotherapy and, in some cohorts, oxaliplatin. Response was classified using the Neoadjuvant Rectal score: <8 low/good response, 8–16 intermediate, >16 high/poor response.
- The NAR score retained its prognostic relevance. Higher NAR was associated with shorter disease-free survival in both major transcriptomic cohorts, with HR 3.04 (95% CI 1.36–6.78; p=0.007) in S:CORT and HR 2.41 (95% CI 1.69–3.42; p<0.01) in GSE87211.
- Seventeen Hallmark pathways were reproducibly enriched in high-NAR tumors. Epithelial-mesenchymal transition was particularly prominent, alongside Wnt/β-catenin, TGF-β, angiogenesis, hypoxia and other stromal/oncogenic programs; low-NAR tumors instead showed enrichment of interferon-α, interferon-γ and allograft-rejection immune signatures.
- Cytotoxic lymphocyte scores were significantly higher in good responders in both transcriptomic cohorts (p=0.045 and p=0.007). In the independent immunofluorescence cohort, 50% of low-NAR tumors versus 25% of high-NAR tumors fell in the highest CD8-density tertile, although this small-cohort difference was not statistically significant.
- Poor responders showed the opposite stromal phenotype. Endothelial-cell scores were higher in high-NAR tumors in both cohorts (p=0.011 and p=0.014), while fibroblast scores were significantly higher in GSE87211 (p=0.027) but not S:CORT (p=0.11). These differences were not explained simply by overall tumor purity.
- Combining fibroblast and endothelial scores relative to cytotoxic lymphocytes improved biological separation but remained inadequate as a classifier. The combined score was higher in high-NAR tumors (p=0.0087 and p=0.00017), yet AUC was only 0.67–0.68 when directly separating high from low NAR and 0.60–0.63 when applied across the full cohorts.
CLINICAL TAKEAWAY
Pretreatment rectal tumors that respond well to neoadjuvant radiation appear biologically more immune-infiltrated, whereas resistant tumors show a stronger stromal/EMT phenotype. The reproducibility across cohorts is compelling mechanistically, but an AUC around 0.6–0.7 is nowhere near sufficient to withhold or intensify radiotherapy based on these biomarkers alone.