Chemoradiotherapy (CRT) and surgical resection is the standard of care for locally advanced carcinoma of the rectum (LARC). Clinical trial comparisons of pre- and post-operative neo-adjuvant CRT indicate that pre-operative therapy has beneficial effects on recurrence rates and acute and long-term toxicity and enabling a higher rate of sphincter saving surgery. There is also concern for patients regarding overtreatment, where the pre-operative therapy is not beneficial. However, it is still difficult to predict which patients will benefit as there are no specific biomarkers available.
The paper by Gonçalves-Ribeiro et al addresses this using transcriptomic profiling of responder and non-responder LARC patients before pre-operative treatment. Pre-treatment biopsies from patients diagnosed with LARC were processed by LCM followed by hybridization to Primeview microarrays. After pre-operative treatment patients' response was determined from pathological response using Mandard classification. Analysis of differential gene expression in tumor tissue showed very little change between responders and non-responders, whereas the cancer-associated stroma had significant transcriptomic changes.
MMP2, fibronectin 1, collagen 1, collagen 3A1 and IGFBP5 were identified as differentially expressed genes that may be amenable to IHC. These genes were confirmed by IHC in a second cohort of rectal cancer tumor samples. IHC was scored and used to predict responders and non-responders in the training set with approx. 68% sensitivity and 81% specificity.
The data was refined further by analyzing how each protein contributed to the discrimination between responder and non-responder. Fibronectin 1 and collagen 3A were the strongest predictors compared to the other three proteins so a risk score was devised using just these two proteins as covariates. This model performed better than the 5- protein model with a 63% sensitivity and a 93% selectivity in the training cohort and 71% sensitivity with 87% selectivity in the validation cohort.
This study identifies:
Most changes in gene expression between responders and non-responders were associated with stromal rather than tumor tissue.
MMP2, fibronectin 1, collagen 1, collagen 3A1 and IGFBP5 were identified from the stroma as potential biomarkers for response to pre-operative treatment.
Developed and validated an IHC score based on two CAF-specific proteins that performed well predicting responders and non-responders
In summary, the authors present additional data confirming a significant effect of cancer associated fibroblasts in the response to therapy with most transcriptomic changes occurring in the stroma and not the tumor tissue. They identified a panel of 5 proteins which could predict response and refined this to CAF-specific proteins that predicted response. This procedure could help stratify patients where pre-treatment could be avoided, these patients may then benefit from advanced surgery or alternative treatment options and prevent unnecessary toxicities.