Blood Test and Tumor Genetics Sharpen Predictions of Rectal Cancer Immunotherapy Success

A new analysis from the TORCH clinical trial has delivered some of the clearest evidence yet that a combination of tumor genetics and serial blood-based biopsy can predict which patients with hard-to-treat rectal cancer will respond to an immunotherapy-centered treatment strategy. The study, published in Genome Medicine, focused on patients with microsatellite stable (MSS) locally advanced rectal cancer, a molecular subtype that has historically resisted the immune checkpoint inhibitors that produce dramatic results in other colorectal cancer subsets. By layering genetic features onto standard imaging assessments, the researchers built predictive models that substantially outperformed imaging alone in identifying complete responders, offering a potential roadmap for personalizing treatment decisions in this challenging disease.

The clinical stakes could hardly be higher. Locally advanced rectal cancer has traditionally been treated with neoadjuvant chemoradiotherapy followed by surgery, an approach that saves lives but often exacts a heavy toll, including permanent colostomies, radiation-related bowel dysfunction, urinary problems, and sexual health consequences. Total neoadjuvant therapy (TNT), which delivers chemotherapy and radiation before any operation, has improved outcomes and opened the door to organ preservation for patients whose tumors vanish completely. The TORCH trial took this concept further by adding immunotherapy to the neoadjuvant regimen, creating immunotherapy-based total neoadjuvant therapy, or iTNT, for MSS patients who would typically be excluded from immunotherapy-based strategies because their tumors lack the microsatellite instability that makes them immunotherapy-sensitive.

In the new study, the research team analyzed clinical and genetic data from 63 patients enrolled in the prospectively registered TORCH trial between May 1, 2021, and September 15, 2022. The investigators collected baseline tissue mutations, magnetic resonance tumor regression grade (mrTRG) scores from treatment-time MRI scans, and circulating tumor DNA (ctDNA) positivity measured at multiple points across the treatment course. Their goal was twofold: first, to determine which features measured before, during, or after iTNT best distinguished patients who achieved a complete response from those who did not, and second, to identify early markers of recurrence risk that could guide follow-up intensity and decisions about whether surgery could be safely avoided.

The genetic findings were striking. A baseline mutation in the SMAD4 gene, a well-known tumor suppressor involved in the TGF-beta signaling pathway that helps regulate cell growth and tissue architecture, was significantly associated with poor response to iTNT, with the association reaching statistical significance at P equals 0.007. Patients carrying SMAD4 mutations were markedly less likely to achieve a complete response, suggesting that this single genetic alteration encodes a fundamental resistance to the combined chemo-radiation-immunotherapy approach. The finding aligns with a growing body of evidence that SMAD4 loss reshapes the tumor microenvironment in ways that blunt immune attack and may drive more aggressive, metastasis-prone biology.

Beyond SMAD4, three additional features correlated with complete response rates: post-treatment mrTRG assessed on MRI (P less than 0.0001), tumor mutational burden, or TMB, which quantifies the number of mutations carried by the tumor (P equals 0.03), and the latest ctDNA status, a measure of whether fragments of tumor DNA were still detectable in the bloodstream after treatment (P equals 0.03). TMB is a classic immunotherapy biomarker, since tumors with more mutations tend to produce more neoantigens that the immune system can recognize, but its relevance in MSS rectal cancer treated with combination iTNT had been uncertain. The ctDNA measurement, meanwhile, represents the concept of a liquid biopsy: rather than sampling the tumor itself, clinicians can track fragments of tumor-derived DNA shed into circulation, gaining a real-time view of residual disease that imaging may miss.

The true advance, however, came when the researchers integrated these features into multivariable models. A model combining post-treatment mrTRG, SMAD4 mutation status, TMB, and latest ctDNA status achieved an area under the curve, or AUC, of 0.92 for discriminating complete responders from non-responders, compared with just 0.75 for a model based on mrTRG alone, a statistically significant improvement (P equals 0.0098). An AUC of 0.92 reflects excellent discriminative accuracy, while 0.75 represents only modest performance, so the jump is clinically meaningful. Importantly, the researchers also tested whether prediction could happen earlier in the treatment course. An early integrated model that used mid-iTNT mrTRG along with the genetic features achieved an AUC of 0.89 versus 0.76 for the mrTRG-only model (P equals 0.047), meaning clinicians might one day gauge a patient’s trajectory partway through therapy rather than waiting until the end.

The prognostic analysis added another layer of clinical utility. Using Cox proportional hazards regression and Kaplan-Meier survival analyses, the team found that baseline SMAD4 mutations and ctDNA positivity at later time points were both correlated with an increased risk of disease progression, including local recurrence and distant metastasis. In practical terms, a patient whose blood still shows molecular traces of tumor after treatment completion, or whose original tumor carried a SMAD4 mutation, faces a higher probability that the cancer will return, even if standard assessments look favorable. This kind of molecular residual disease detection is becoming one of the most actively pursued frontiers in oncology, because it can flag recurrence months or even years before radiographic evidence appears, when intervention may be most effective.

The implications for treatment strategy are substantial. For patients with locally advanced rectal cancer, the decision of whether to proceed to surgery or pursue nonoperative management, sometimes called watch and wait, hinges on the confidence with which a complete response can be established. Current assessments rely on a patchwork of digital rectal examination, endoscopy, MRI, and serum markers such as carcinoembryonic antigen, none of which is perfectly reliable. A molecular framework that adds SMAD4 status and serial ctDNA to the mix could give clinicians the confidence to safely spare surgery in true complete responders, while identifying poor responders early enough to intensify their therapy, enroll them in alternative trials, or avoid prolonging an ineffective course of immunotherapy. The study’s authors emphasize that these models are best understood as complementary risk stratification tools rather than standalone decision instruments, augmenting rather than replacing clinical judgment and imaging.

It is worth underscoring what makes the MSS population such an important target. Roughly the overwhelming majority of colorectal cancers are microsatellite stable, and for years, immune checkpoint inhibitors were considered largely futile in this group. The TORCH trial’s demonstration that adding immunotherapy to total neoadjuvant therapy can produce promising complete response rates in MSS patients was itself a paradigm-shifting result, and the current analysis addresses the critical next question: how do we know who is benefiting? Without reliable biomarkers, the field risks either overtreating patients who will never respond or undertreating those who would have responded with more time. The integrated models described here offer a data-driven way to navigate that uncertainty, converting a one-size-fits-all regimen into something closer to adaptive, biology-informed care.

Limitations remain, as they do in any single-trial biomarker analysis of 63 patients. The findings will need validation in larger, independent cohorts, and the logistics of serial ctDNA monitoring, including assay standardization, cost, and the optimal sampling schedule, remain active areas of investigation. The trial was registered as NCT04518280, and the analysis was funded by the National Natural Science Foundation of China and the Beijing Xisike Clinical Oncology Research Foundation. Still, the convergence of tissue genomics, liquid biopsy, and quantitative imaging in a prospectively registered trial represents a template for how biomarker-driven personalization should be developed in rectal cancer. If validated, the combination of SMAD4 mutation status, tumor mutational burden, serial ctDNA tracking, and mrTRG assessment could become a standard part of the decision-making toolkit for MSS locally advanced rectal cancer, helping more patients keep their organs, avoid futile toxicity, and face their disease with a clearer, molecularly grounded picture of what lies ahead.

Subject of Research: Predicting response and prognosis of immunotherapy-based total neoadjuvant therapy in microsatellite stable locally advanced rectal cancer using genetic features and serial circulating tumor DNA

Article Title: Application of genetic features and serial ctDNA in estimating response and prognosis of immunotherapy-based total neoadjuvant therapy (iTNT) for microsatellite stable locally advanced rectal cancer: data from TORCH trial

Article References: Wang, Y., Xu, Y., Lin, Y., Liu, Y., Jing, Q., Shen, L., Wan, J., Wang, Y., Zhang, H., Wu, R., Zhou, S., Chen, Y., Zhang, Z., Wang, J., Sun, Y., Bao, H., Ou, Q., Bao, H., Shao, Y., … Xia, F. (2026). Application of genetic features and serial ctDNA in estimating response and prognosis of immunotherapy-based total neoadjuvant therapy (iTNT) for microsatellite stable locally advanced rectal cancer: data from TORCH trial. Genome Medicine. https://doi.org/10.1186/s13073-026-01755-2

Image Credits: AI Generated

DOI: 10.1186/s13073-026-01755-2

Keywords: rectal cancer, immunotherapy, ctDNA, SMAD4 mutation, tumor mutational burden, total neoadjuvant therapy, microsatellite stable, TORCH trial, liquid biopsy, organ preservation, prognosis, complete response