Every two weeks, a critical review of the most significant colorectal cancer research indexed in PubMed, with key findings, methodological appraisal, and clinical implications.
1. Circulating tumor DNA status and adjuvant chemotherapy after resection of colorectal liver metastases (GALAXY/CIRCULATE-Japan, JAMA Oncology)
Kozo Kataoka (Division of Lower GI Surgery, Department of Gastroenterological Surgery, Hyogo Medical University, Nishinomiya, Japan), K. Ito, Y. Nakamura, et al., for the CIRCULATE-Japan Group
Key points
- Prognostic analysis of 298 patients with resected colorectal liver metastases (CRLM) and no extrahepatic disease from the GALAXY arm of CIRCULATE-Japan (UMIN000039205): 191 had upfront surgery and 107 neoadjuvant chemotherapy; median follow-up 43.2 months.
- Molecular residual disease (MRD) was assessed 2 to 10 weeks after hepatectomy with the tumor-informed Signatera assay; 32.4% (62/191) of upfront-surgery patients and 47.7% (51/107) of neoadjuvant-treated patients were MRD-positive.
- In the upfront-surgery cohort, MRD positivity predicted worse DFS (HR 4.14; 95% CI 2.69-6.36) and OS (HR 9.13; 95% CI 4.44-18.78); 48-month OS was 41.4% vs 91.1%.
- Among MRD-positive upfront-surgery patients, adjuvant chemotherapy was associated with longer OS (adjusted HR 0.27; 95% CI 0.08-0.88; P=.03; 48-month OS 65.3% vs 32.9%) and DFS (adjusted HR 0.07; 95% CI 0.02-0.24); no significant benefit was seen in MRD-negative patients (OS adjusted HR 0.54; 95% CI 0.10-2.88).
- In the neoadjuvant cohort, adjuvant chemotherapy was not associated with benefit regardless of MRD status (MRD-positive OS HR 0.81; 95% CI 0.27-2.41).
Key limitations
- Observational, nonrandomized comparison of adjuvant chemotherapy versus observation; treatment selection bias is likely despite adjustment.
- Subgroups are small and confidence intervals wide; the authors describe the predictive analyses as underpowered.
- The journal has posted a correction to the abstract, key points, text, one figure and a supplement; some reported values may differ between versions.
Future directions
- Prospective randomized trials of MRD-guided adjuvant chemotherapy after upfront hepatectomy.
- Biologically informed strategies, including targeted agents, for patients with persistent MRD after neoadjuvant chemotherapy.
- Use of serial MRD surveillance and ctDNA levels to tailor imaging and treatment intensity.
Hepatic resection offers the possibility of cure for selected patients with colorectal liver metastases, but a large share of patients relapse after surgery, most often in the remnant liver or the lungs. Perioperative or adjuvant systemic chemotherapy is widely used to reduce this risk, yet randomized evidence for an overall survival benefit of chemotherapy around hepatectomy has been inconsistent, and clinicians have no reliable way to identify which patients carry residual micrometastatic disease and which are already cured by surgery.
Circulating tumor DNA detected after curative-intent surgery has emerged as a highly specific marker of molecular residual disease (MRD) in colorectal cancer, and the GALAXY study, the prospective observational arm of the nationwide CIRCULATE-Japan platform, previously showed in Nature Medicine that postoperative ctDNA strongly predicts recurrence. Whether MRD status can also indicate who benefits from adjuvant chemotherapy after liver resection, the setting in which treatment decisions are most uncertain, had not been examined in a large cohort with mature survival follow-up.
Kataoka and colleagues analysed 298 GALAXY patients enrolled between May 2020 and July 2024 who had resected CRLM without extrahepatic disease and were not enrolled in interventional trials within the platform. MRD was assessed 2 to 10 weeks after hepatectomy with Signatera, a personalized, tumor-informed multiplex PCR-NGS assay, with positivity defined as two or more tumor-specific variants above threshold; blood was then drawn serially up to 96 weeks. The upfront-surgery (n=191) and neoadjuvant chemotherapy (n=107) cohorts were analysed separately, with Cox models and 70-day landmark analyses to address immortal time bias. Median age was 67 years and median follow-up 43.2 months.
Postoperative MRD positivity was the strongest independent adverse factor for both endpoints. In the upfront-surgery cohort, 32.4% were MRD-positive, with a DFS hazard ratio of 4.14 (95% CI 2.69-6.36) and an OS hazard ratio of 9.13 (95% CI 4.44-18.78); 48-month OS was 41.4% versus 91.1%. In multivariable analysis the MRD hazard ratio was 7.89 for DFS and 12.03 for OS. Among MRD-positive upfront-surgery patients, adjuvant chemotherapy was associated with markedly better outcomes: adjusted OS HR 0.27 (95% CI 0.08-0.88; P=.03), with 48-month OS of 65.3% versus 32.9%, and adjusted DFS HR 0.07 (95% CI 0.02-0.24; P<.001).
No significant benefit of adjuvant chemotherapy was found in MRD-negative upfront-surgery patients (DFS adjusted HR 0.69; 95% CI 0.32-1.50; OS adjusted HR 0.54; 95% CI 0.10-2.88). After neoadjuvant chemotherapy, nearly half of patients (47.7%) were MRD-positive, and adjuvant chemotherapy showed no association with benefit in either MRD-positive (DFS HR 0.69; OS HR 0.81) or MRD-negative patients. Patients who remained serially MRD-negative had far better DFS and OS than those MRD-positive at any timepoint. Liver recurrence predominated in MRD-positive patients, whereas lung recurrence was more frequent among MRD-negative patients.
The central limitation is that adjuvant chemotherapy was not randomized: patients who received it may have differed in fitness, tumor burden or surgeon preference, and adjusted hazard ratios cannot remove all such bias. The authors acknowledge that the subgroup analyses were underpowered and that predictive value must be confirmed in prospective randomized trials; some confidence intervals are extremely wide. Sample sizes differ slightly between the prognostic and predictive analyses in secondary reports, and the journal has since posted a correction affecting the abstract, key points, text, Figure 4 and a supplement, so readers should consult the corrected version for exact values.
Even with these caveats, GALAXY provides the most detailed evidence to date that postoperative ctDNA identifies a group of patients with resected liver metastases at very high risk of death, and suggests that adjuvant chemotherapy may matter most in that group after upfront surgery. The lack of benefit after neoadjuvant chemotherapy suggests that persistent MRD in this setting may reflect chemoresistant disease, for which intensifying cytotoxic therapy alone may be insufficient. The authors name prospective validation in upfront-surgery patients as the priority next step. The results were also presented orally at the ESMO Gastrointestinal Cancers Congress 2026.
Source: Kataoka et al., "Circulating Tumor DNA Status and Adjuvant Chemotherapy in Resected Colorectal Liver Metastases," JAMA Oncology, published online July 3, 2026. Read the paper
2. Second-line adagrasib plus cetuximab versus chemotherapy in KRAS G12C-mutated metastatic colorectal cancer (KRYSTAL-10, ESMO GI 2026)
Presenting author Josep Tabernero (Vall d'Hebron University Hospital, Barcelona, Spain), with S. Kopetz, J. Lee, et al., on behalf of the KRYSTAL-10 investigators
Key points
- Global, open-label, randomized phase 3 trial (NCT04793958) of 461 patients with KRAS G12C-mutated metastatic colorectal cancer progressing after first-line fluoropyrimidine-based doublet chemotherapy; adagrasib 600 mg twice daily plus cetuximab (n=231) versus FOLFIRI or mFOLFOX6 with or without a VEGF/VEGFR inhibitor (n=230).
- Dual primary endpoints were not met: median PFS by blinded independent central review 7.5 vs 8.1 months (HR 0.89; 95% CI 0.71-1.13; P=.32); median OS 21.6 vs 21.7 months (HR 0.83; 95% CI 0.67-1.03; P=.09).
- Objective response rate was higher with adagrasib plus cetuximab: 47% (95% CI 40%-53%) vs 16% (95% CI 11%-21%); complete responses 7% vs less than 1%.
- Grade 3-5 treatment-emergent adverse events: 46% vs 55%; serious events 9% vs 12%; no new safety signals.
- Subsequent KRAS G12C inhibitor use was 4% vs 30%, which may have affected OS. Presented as LBA1 at ESMO GI 2026 (Munich, July 1-4); not yet published in a peer-reviewed journal.
Key limitations
- Open-label design with informative censoring of PFS from subsequent therapies started before central confirmation of progression.
- Crossover-like effect: about 30% of chemotherapy-arm patients later received a KRAS G12C inhibitor, potentially diluting any OS difference.
- Conference data only; full publication with subgroup, duration-of-response and quality-of-life data is awaited.
Future directions
- Regulatory review of the 2024 US accelerated approval, for which KRYSTAL-10 was the confirmatory trial.
- Evaluation of KRAS G12C inhibitor combinations earlier in the disease course and with chemotherapy backbones.
- Use of the high response rate in settings where tumor shrinkage matters, such as potentially resectable disease.
KRAS G12C mutations define a small molecular subset of metastatic colorectal cancer that has historically carried a poor prognosis and responds poorly to standard therapy. Unlike in lung cancer, KRAS G12C inhibitors given alone have limited activity in colorectal cancer because of feedback reactivation of EGFR signalling, which provided the rationale for combining the inhibitor adagrasib with the anti-EGFR antibody cetuximab. In the second-line setting, the alternative for these patients has been a standard chemotherapy doublet with or without an anti-angiogenic agent, the comparator chosen for KRYSTAL-10.
In the single-arm KRYSTAL-1 study, published in the New England Journal of Medicine in 2023, adagrasib plus cetuximab produced an objective response rate of 46% (95% CI 28%-66%) among 28 evaluable heavily pretreated patients, with a median duration of response of 7.6 months and median PFS of 6.9 months. On this basis the US Food and Drug Administration granted accelerated approval in 2024 for previously treated KRAS G12C-mutated colorectal cancer, contingent on a confirmatory randomized trial. KRYSTAL-10 was designed as that trial, testing the combination against standard chemotherapy in the second-line setting.
KRYSTAL-10 randomized 461 patients 1:1 after progression on first-line fluoropyrimidine-based doublet chemotherapy with oxaliplatin or irinotecan. The experimental arm received adagrasib 600 mg twice daily plus cetuximab 500 mg/m² every two weeks; the control arm received investigator's choice of FOLFIRI or mFOLFOX6 with or without a VEGF or VEGFR inhibitor. The dual primary endpoints were PFS by blinded independent central review and OS, with objective response rate and safety as key secondary endpoints. The minimum follow-up for OS was 26.4 months.
Neither primary endpoint reached statistical significance. Median PFS was 7.5 months with adagrasib plus cetuximab and 8.1 months with chemotherapy (HR 0.89; 95% CI 0.71-1.13; P=.32). Median OS was 21.6 versus 21.7 months (HR 0.83; 95% CI 0.67-1.03; P=.09). The combination did, however, produce a substantially higher objective response rate, 47% (95% CI 40%-53%) versus 16% (95% CI 11%-21%), including complete responses in 7% versus less than 1%.
Tolerability favored the targeted combination. Any-grade treatment-emergent adverse events occurred in 98% versus 96% of treated patients, grade 3-5 events in 46% versus 55%, and serious events in 9% versus 12%; discontinuation for treatment-related toxicity was uncommon in both arms. One treatment-related death occurred in the chemotherapy arm, and no new safety signals were identified. Post-progression therapy differed markedly: 30% of patients in the chemotherapy arm went on to receive a KRAS G12C inhibitor, compared with 4% in the experimental arm.
Several factors complicate interpretation. PFS assessment was affected by informative censoring when patients started subsequent therapies before progression was confirmed centrally, and the open-label design may have influenced treatment decisions. The subsequent use of KRAS G12C inhibitors in the control arm may have narrowed any OS difference, although the PFS result does not depend on this. Commentators described the outcome as disappointing given the high response rate. Results are available only from the congress presentation and secondary coverage and have not yet appeared in a peer-reviewed journal.
KRYSTAL-10 is a negative confirmatory trial: a regimen that markedly increases tumor shrinkage did not prolong PFS or OS compared with standard second-line chemotherapy. Because it was the confirmatory study for the 2024 accelerated approval, that approval is now at risk of withdrawal. The trial also illustrates the limits of response rate as a surrogate endpoint in colorectal cancer. Future work is likely to focus on next-generation KRAS inhibitors, combinations with chemotherapy, and settings in which deep responses carry direct clinical value.
Source: KRYSTAL-10 trial, presented at the ESMO Gastrointestinal Cancers Congress 2026 (abstract LBA1), Munich, July 2026; not yet published in a peer-reviewed journal. Conference coverage (Targeted Oncology)
3. Duration of adjuvant mFOLFOX6 and atezolizumab in stage III dMMR colon cancer: exploratory landmark analysis (ATOMIC, ESMO GI 2026)
Presenting author Anke Reinacher-Schick (Clinic for Hematology and Oncology with Palliative Medicine, Katholisches Klinikum Bochum, Germany), on behalf of the ATOMIC (A021502/AIO-KRK-0317) investigators
Key points
- Retrospective, exploratory post hoc analysis of the phase 3 ATOMIC trial (NCT02912559), which randomized patients with stage III mismatch repair-deficient (dMMR) colon cancer to mFOLFOX6 plus atezolizumab (then 6 months of atezolizumab alone) or mFOLFOX6 alone; 611 patients in the FOLFOX-duration landmark analysis.
- With more than 6 FOLFOX cycles, adding atezolizumab was associated with improved 3-year DFS: 88.2% vs 79.1% (adjusted HR 0.45; 95% CI 0.29-0.71).
- With 6 or fewer FOLFOX cycles, no benefit was seen: 3-year DFS 80.0% vs 79.9% (adjusted HR 0.84; 95% CI 0.35-2.01).
- Within the atezolizumab arm, 12 or more versus fewer than 12 atezolizumab cycles: 3-year DFS 88.0% vs 84.0% (adjusted HR 0.81; 95% CI 0.38-1.71; P=.58).
- Atezolizumab was continued after FOLFOX stopped in 87% of patients receiving more than 6 cycles vs 48% with 6 or fewer. Presented as abstract 1O at ESMO GI 2026; not yet published in a peer-reviewed journal.
Key limitations
- Retrospective, post hoc and hypothesis-generating; treatment duration was not randomized.
- Duration reflected toxicity and patient or investigator choice, so healthier patients may have received more treatment (confounding by fitness).
- Small subgroups with wide confidence intervals, and inconsistent subgroup sizes in secondary reports.
Future directions
- Prospective trials of shorter or de-escalated chemotherapy backbones with PD-L1 or PD-1 blockade in dMMR colon cancer.
- Neoadjuvant immunotherapy strategies that may avoid adjuvant chemotherapy altogether in dMMR disease.
- Biomarkers such as ctDNA to guide how long immunotherapy is needed.
A minority of stage III colon cancers are mismatch repair-deficient (dMMR). Although dMMR tumors have a relatively favorable prognosis overall, they respond less consistently to fluoropyrimidine-based therapy, and their high neoantigen load makes them highly sensitive to immune checkpoint blockade. Adjuvant FOLFOX remains the backbone of postoperative treatment for stage III disease, and the question of how immunotherapy should be integrated with it has become central to this subgroup.
The phase 3 ATOMIC trial (A021502/AIO-KRK-0317), whose primary results appeared in the New England Journal of Medicine, showed that adding the anti-PD-L1 antibody atezolizumab to 6 months of adjuvant mFOLFOX6, followed by 6 months of atezolizumab monotherapy, improved disease-free survival in stage III dMMR colon cancer. In routine practice, however, many patients cannot complete 12 cycles of FOLFOX because of neuropathy or other toxicity, and the optimal duration of both chemotherapy and atezolizumab has not been defined. This exploratory analysis asked whether the benefit of atezolizumab depends on how much treatment patients actually receive.
The investigators performed a landmark analysis at 12 cycles (6 months) among patients who had received at least one cycle and were alive and free of progression at that point. Patients were grouped by the number of mFOLFOX6 cycles received (6 or fewer versus more than 6) and, in the combination arm, by the number of atezolizumab cycles (fewer than 12 versus 12 or more). Groups were compared using log-rank tests, with hazard ratios from Cox models adjusted for T and N stage. The FOLFOX-duration analysis included 611 patients; 55.0% were female and 62.5% had N1 or N1c disease.
The benefit of atezolizumab was concentrated in patients who received more than 6 cycles of FOLFOX. In this group, 3-year DFS was 88.2% with the combination versus 79.1% with FOLFOX alone (adjusted HR 0.45; 95% CI 0.29-0.71). Among patients who received 6 or fewer cycles, 3-year DFS was essentially identical, 80.0% versus 79.9% (adjusted HR 0.84; 95% CI 0.35-2.01). Compared with FOLFOX alone, 12 or more atezolizumab cycles was associated with an adjusted HR of 0.48 (95% CI 0.31-0.74), whereas fewer than 12 cycles gave an HR of 0.59 (95% CI 0.29-1.18).
Within the combination arm, receiving 12 or more cycles of atezolizumab was associated with numerically higher 3-year DFS than fewer cycles (88.0% vs 84.0%; adjusted HR 0.81; 95% CI 0.38-1.71; P=.58), a difference that was not statistically significant. Each additional FOLFOX cycle was associated with a hazard ratio of 0.93 (95% CI 0.84-1.03) in the combination arm and 0.98 (95% CI 0.91-1.05) with FOLFOX alone. Notably, atezolizumab was continued after FOLFOX ended in 87% of patients who received more than 6 FOLFOX cycles but in only 48% of those who received 6 or fewer, so shorter chemotherapy was often a marker of shorter immunotherapy as well.
The analysis is retrospective, post hoc and explicitly hypothesis-generating. Treatment duration was determined by toxicity and patient or investigator choice rather than randomization, so patients who completed more treatment were likely fitter or had more favorable disease, a form of confounding that adjustment for T and N stage cannot fully address. The landmark design excludes patients who relapsed or died within 6 months, several key comparisons have confidence intervals crossing 1, and secondary reports give inconsistent subgroup sizes. The findings were presented at a congress and have not yet been peer reviewed.
For clinicians, the analysis gives no support for deliberately shortening adjuvant therapy in dMMR colon cancer; the presenter concluded that the optimal duration remains unknown and that patients should continue treatment for as long as they tolerate it. The pattern also raises the question of whether the observed benefit reflects chemotherapy, immunotherapy or their combined duration. Prospective trials testing de-escalated chemotherapy with checkpoint inhibition, and neoadjuvant immunotherapy approaches in dMMR colon cancer, are better placed to answer these questions.
Source: ATOMIC trial exploratory analysis, presented at the ESMO Gastrointestinal Cancers Congress 2026 (abstract 1O), Munich, July 2026; not yet published in a peer-reviewed journal. Conference coverage (OncLive)
4. Plasma ctDNA hyperselection to guide continued cetuximab across first- and second-line therapy in RAS/BRAF wild-type metastatic colorectal cancer (CAPRI-2 GOIM, ESMO GI 2026)
Presenting author Giulia Martini (Università degli Studi della Campania Luigi Vanvitelli, Naples, Italy), on behalf of the CAPRI-2 GOIM investigators
Key points
- Biomarker-driven phase 2 trial (NCT05312398) in RAS and BRAF V600E wild-type, microsatellite-stable metastatic colorectal cancer: first-line FOLFIRI plus cetuximab, then FOLFOX plus cetuximab at progression, with plasma ctDNA genomic profiling at baseline and at each progression.
- Of 182 first-line patients, 132 (73%) were "negative hyperselected" (no ctDNA resistance variants in genes such as KRAS, NRAS, BRAF, PIK3CA, HER2 and the EGFR extracellular domain) and 50 (27%) were positive hyperselected.
- First line, negative vs positive hyperselected: ORR 78.0% vs 54.0% (OR 3.00; 95% CI 1.42-6.38; P=.003); median PFS 13.8 vs 9.6 months (HR 0.55; 95% CI 0.39-0.78; P=.001).
- Second line (n=75): median PFS 8.3 vs 4.65 months (HR 0.43; 95% CI 0.25-0.75; P=.003); ORR 27.8% vs 20.0% (P=.565, not significant); OS from second-line start 27.4 vs 15.9 months (HR 0.60; 95% CI 0.29-1.24; P=.170).
- Presented as abstract 2RO at ESMO GI 2026; not yet published in a peer-reviewed journal.
Key limitations
- Single-arm design: all patients received cetuximab, so the analysis compares biomarker groups rather than treatments.
- Second-line cohort is small and the split by hyperselection status was not fully reported; OS from second-line start was not significant.
- Safety data and full methods are not yet available outside the congress presentation.
Future directions
- Randomized testing of ctDNA-guided anti-EGFR continuation versus switching biologic at first progression.
- Report of the planned third-line cetuximab-based phase of CAPRI-2.
- Harmonized ctDNA panels and thresholds for defining hyperselection in routine practice.
Anti-EGFR monoclonal antibodies such as cetuximab are a standard first-line partner for chemotherapy in RAS wild-type metastatic colorectal cancer, particularly in left-sided tumors. Not all RAS wild-type tumors respond, however: primary resistance can arise from alterations in other genes in the EGFR signalling pathway, and acquired resistance emerges under treatment pressure. Tissue testing at diagnosis captures only part of this heterogeneity, and it is rarely repeated at progression.
The concept of "hyperselection" extends molecular screening beyond RAS and BRAF to a broader set of resistance genes, and the Italian GOIM group has previously studied liquid biopsy biomarkers for cetuximab-based rechallenge in its CAVE studies. Plasma ctDNA profiling offers a minimally invasive way to define hyperselection at baseline and to follow clonal evolution at each progression. CAPRI-2 GOIM was designed to test whether this approach can guide cetuximab-based therapy across successive lines in a molecularly selected population.
CAPRI-2 GOIM enrolled patients with RAS and BRAF V600E wild-type, microsatellite-stable metastatic colorectal cancer to receive cetuximab across up to three sequential lines; this report covered first-line FOLFIRI plus cetuximab and second-line FOLFOX plus cetuximab. Comprehensive plasma ctDNA genomic profiling was performed at baseline and at first and second progression. Patients were classified as negative hyperselected when no pathogenic variants were detected in anti-EGFR resistance genes, including KRAS, NRAS, BRAF, PTEN, MAP2K1, NF1, FGFR, ROS1, MET, HER2, the EGFR extracellular domain, PIK3CA, RET and NTRK, and as positive hyperselected when one or more were found.
Of 182 patients treated in the first line, 132 (73%) were negative hyperselected. These patients had a higher objective response rate than positive hyperselected patients (78.0% vs 54.0%; OR 3.00; 95% CI 1.42-6.38; P=.003) and longer median PFS (13.8 vs 9.6 months; HR 0.55; 95% CI 0.39-0.78; P=.001). In the second line, 75 patients received FOLFOX plus cetuximab, and negative hyperselection was again associated with longer median PFS, 8.3 versus 4.65 months (HR 0.43; 95% CI 0.25-0.75; P=.003).
Other second-line endpoints were less conclusive. The objective response rate was 27.8% versus 20.0% (OR 1.53; 95% CI 0.40-7.31; P=.565). Median OS measured from baseline was not reached in negative hyperselected patients versus 22.4 months in positive hyperselected patients (HR 0.47; 95% CI 0.30-0.73; P=.001), but OS measured from the start of second-line therapy, 27.4 versus 15.9 months, did not reach significance (HR 0.60; 95% CI 0.29-1.24; P=.170). Safety data were not reported in the available coverage.
Because every patient received cetuximab, the trial cannot show that continuing cetuximab is better than switching to a different biologic; it shows that ctDNA hyperselection identifies patients who do better on cetuximab-based therapy, which may be partly prognostic rather than predictive. The second-line cohort is small, the number of negative and positive hyperselected patients within it was not reported in the coverage reviewed, and several comparisons were not significant. The analysis rests on a specific ctDNA panel whose findings may not transfer to other assays, and the data have not yet been peer reviewed.
CAPRI-2 GOIM supports the feasibility of serial plasma profiling to guide anti-EGFR therapy over the course of the disease, and suggests that continuing cetuximab while switching the chemotherapy backbone may be reasonable in negative hyperselected patients at first progression. The authors suggest serial liquid biopsy could guide anti-EGFR continuation and sequencing. Randomized trials comparing ctDNA-guided continuation with standard switching, and results from the planned later-line phase, are needed before this becomes routine practice.
Source: CAPRI-2 GOIM trial, presented at the ESMO Gastrointestinal Cancers Congress 2026 (abstract 2RO), Munich, July 2026; not yet published in a peer-reviewed journal. Conference coverage (OncoDaily)
5. Short-course radiotherapy added to CAPOX plus toripalimab in pMMR locally advanced rectal cancer (TORCH-iTNT, ESMO GI 2026)
Presenting author Zhen Zhang (Fudan University Shanghai Cancer Center, Shanghai, China), on behalf of the TORCH-iTNT investigators
Key points
- Multicenter, randomized phase 2 trial (NCT06281405) of 192 patients with mismatch repair-proficient (pMMR) T3-4 and/or node-positive, nonmetastatic rectal cancer.
- Group A: six cycles of CAPOX plus the anti-PD-1 antibody toripalimab; group B: short-course radiotherapy (25 Gy in 5 fractions) followed by the same six cycles.
- Among 158 patients who had completed treatment by March 20, 2026 (96.2% with high-risk features), overall complete response (pCR or clinical complete response under watch-and-wait) was 60.2% with radiotherapy vs 37.3% without, an absolute difference of 22.9 percentage points.
- Clinical complete responses managed by watch-and-wait: 27/83 vs 9/75; pCR among operated patients: 43.4% (23/53) vs 32.8% (19/58).
- Grade 3-4 thrombocytopenia, the most common grade 3-4 adverse event, occurred in 14.5% vs 13.3%. Presented at ESMO GI 2026; not yet published in a peer-reviewed journal.
Key limitations
- Randomized phase 2 with an interim-style report of 158 of 192 patients; no odds ratios, confidence intervals or P values were reported.
- Complete response combines pathologic and clinical assessments, and clinical complete response rests on imaging and endoscopy without long-term regrowth data.
- Single-country trial using a PD-1 antibody not widely available outside China; survival, functional and quality-of-life outcomes are not yet reported.
Future directions
- Mature data on organ preservation, local regrowth, anorectal function and survival.
- Phase 3 confirmation and comparison with standard total neoadjuvant therapy without immunotherapy.
- Biomarkers to identify pMMR patients most likely to benefit from radiotherapy-primed immunotherapy.
Total neoadjuvant therapy, combining radiotherapy and systemic chemotherapy before surgery, has become a standard approach for locally advanced rectal cancer and has increased complete response rates, opening the door to nonoperative management. Immune checkpoint blockade is highly effective in mismatch repair-deficient rectal tumors, but most rectal cancers are mismatch repair-proficient (pMMR) and are largely resistant to PD-1 blockade on its own. For these patients, finding a way to make immunotherapy work is seen as one of the main routes to higher complete response rates and more organ preservation.
Radiotherapy may make pMMR tumors more visible to the immune system by inducing immunogenic cell death and remodelling the tumor microenvironment, and short-course regimens are thought to be particularly suitable partners for immunotherapy. Several Chinese studies have combined short-course radiotherapy, chemotherapy and PD-1 inhibitors in pMMR rectal cancer with encouraging response rates. What remained unclear was how much radiotherapy itself contributes when chemotherapy and a PD-1 antibody are already given, a question TORCH-iTNT addressed directly by randomizing the radiotherapy component.
TORCH-iTNT is a multicenter, randomized phase 2 trial, sponsored by Fudan University, in patients with pMMR, T3-4 and/or node-positive nonmetastatic rectal cancer. Patients were randomized to six cycles of CAPOX plus toripalimab (group A) or short-course radiotherapy, 25 Gy in 5 fractions, followed by the same immunochemotherapy (group B). The primary endpoint was overall complete response, defined as pathologic complete response after surgery or clinical complete response in patients managed by watch-and-wait. Secondary endpoints included organ preservation, anorectal function, adverse events and survival. A total of 192 patients were randomized.
By March 20, 2026, 158 patients had completed treatment (75 in group A and 83 in group B), and 96.2% had high-risk features. The overall complete response rate was 60.2% with short-course radiotherapy versus 37.3% with immunochemotherapy alone, an absolute difference of 22.9 percentage points. Clinical complete responses leading to watch-and-wait management occurred in 27 of 83 patients in group B versus 9 of 75 in group A. Among patients who underwent surgery, the pathologic complete response rate was 43.4% (23/53) versus 32.8% (19/58).
Toxicity was similar between arms. The most common grade 3-4 adverse event was thrombocytopenia, in 14.5% of group B and 13.3% of group A. Progression during treatment was rare (one patient in group B and two in group A). Six patients in group A and two in group B without a complete response did not proceed to surgery because they declined or received further treatment. Data on organ preservation over time, anorectal function and survival were not reported in the available coverage.
The results are preliminary. The report covers 158 of 192 randomized patients, and no odds ratio, confidence interval or P value for the primary comparison was given in the coverage reviewed, so the statistical robustness of the difference cannot be judged. The composite endpoint mixes pathologic and clinical assessments; clinical complete response depends on imaging and endoscopy, and without follow-up for local regrowth it is not equivalent to cure. The trial was conducted in China with toripalimab, and the results have not yet been peer reviewed.
If confirmed, TORCH-iTNT would indicate that short-course radiotherapy contributes substantially to complete response when combined with chemotherapy and PD-1 blockade in pMMR rectal cancer, and that immunochemotherapy alone is not a sufficient substitute for radiotherapy in this population. Alongside randomized evidence such as the phase 3 UNION trial, it strengthens the case for radiotherapy-primed immunotherapy as a route to organ preservation. Mature organ-preservation, functional and survival data, and phase 3 confirmation, are the necessary next steps.
Source: TORCH-iTNT trial, presented at the ESMO Gastrointestinal Cancers Congress 2026 (oral session), Munich, July 2026; not yet published in a peer-reviewed journal. Conference coverage (OncoDaily)
6. Botensilimab plus balstilimab in refractory microsatellite-stable metastatic colorectal cancer without active liver metastases: 3-year survival (C-800-01, ESMO GI 2026)
Presenting author Benjamin L. Schlechter (Dana-Farber Cancer Institute), on behalf of the C-800-01 investigators
Key points
- Phase 1b first-in-human study (NCT03860272); this cohort comprised 123 patients with refractory microsatellite-stable (MSS) metastatic colorectal cancer without active liver metastases, with a median of 3 prior lines of therapy.
- Botensilimab (anti-CTLA-4) 1 or 2 mg/kg every 6 weeks plus the anti-PD-1 antibody balstilimab 3 mg/kg every 2 weeks.
- Median OS 21.2 months; OS 41% at 24 months and 33% at 3 years, with a plateau in the survival curve beyond 2 years.
- Confirmed objective response rate 21% (3 complete, 23 partial responses); median duration of response not reached; 17% of patients were alive and off all systemic therapy at last follow-up.
- Immune-mediated diarrhea/colitis in 42% (grade 3 or higher 15%), lower at the 1 mg/kg dose (27%; grade 3 or higher 10%); no treatment-related deaths. Presented at ESMO GI 2026; not yet published in a peer-reviewed journal.
Key limitations
- Single-arm phase 1b cohort with no comparator; cross-trial comparison with historical later-line outcomes is unreliable.
- Highly selected population excluding patients with active liver metastases, who make up a large share of MSS metastatic disease.
- OS was an exploratory endpoint, and the FDA has advised that response rate may not reliably predict survival benefit for this regimen.
Future directions
- Results of the randomized phase 3 BATTMAN trial (NCT07152821) of botensilimab plus balstilimab versus best supportive care.
- Mechanistic work on why active liver metastases are associated with resistance to immunotherapy in MSS disease.
- Strategies to extend benefit to patients with liver metastases, such as liver-directed therapy combinations.
Microsatellite-stable (MSS), mismatch repair-proficient tumors account for the large majority of metastatic colorectal cancers and have been almost uniformly resistant to immune checkpoint inhibitors. Patients with refractory MSS disease who have exhausted standard chemotherapy and biologics rely on later-line options such as regorafenib, trifluridine/tipiracil with or without bevacizumab, and fruquintinib, which extend survival only modestly; historical later-line standards in this population have a median OS of roughly 10 to 14 months. Bringing durable immunotherapy benefit to MSS disease is one of the central goals of colorectal cancer research.
Botensilimab is a next-generation anti-CTLA-4 antibody and balstilimab an anti-PD-1 antibody. The first-in-human C-800-01 study has focused its colorectal analyses on patients without active liver metastases, reflecting the long-standing observation that liver involvement is associated with poor outcomes on checkpoint inhibitors in MSS disease. A randomized phase 2 study (NCT05608044) reported an interim response rate of 19.4% with botensilimab 75 mg plus balstilimab. This update provides long-term survival data from the phase 1b cohort.
The analysis included 123 patients with MSS metastatic colorectal cancer without active liver metastases treated in the phase 1b C-800-01 study (NCT03860272). Patients received botensilimab 1 mg/kg or 2 mg/kg every 6 weeks plus balstilimab 3 mg/kg every 2 weeks; 1 mg/kg is the dose selected for phase 3. Patients had received a median of 3 prior lines, 67% had received at least 3, 15% had prior anti-PD-1 or PD-L1 therapy with or without anti-CTLA-4, and 30% had received at least one late-line regimen. The primary endpoint was safety and tolerability; response, disease control and PFS were secondary, and OS was exploratory.
With extended follow-up, median OS was 21.2 months. The OS rate was 41% at 24 months and 33% at 3 years, and the Kaplan-Meier curve plateaued beyond 2 years. The confirmed objective response rate was 21%, comprising 3 complete and 23 partial responses, and the median duration of response had not been reached, with responses lasting from 1.9 months to at least 37.4 months. The disease control rate at 6 weeks was 69%, and 17% of patients (21) were alive and off all systemic anticancer therapy at last follow-up, including 13 responders.
In a post hoc analysis of 37 patients who had received at least one late-line agent (regorafenib, trifluridine/tipiracil with or without bevacizumab, or fruquintinib), the response rate was 22%, median OS 16.2 months and 3-year OS 30%. Immune-mediated diarrhea or colitis was the most common immune-related adverse event, occurring in 42% of patients overall (grade 3 or higher, 15%), and resolved in 98% with a median time to resolution of 14 days. At the 1 mg/kg botensilimab dose the rate was 27% (grade 3 or higher, 10%). There were no treatment-related deaths and no new safety signals.
These are single-arm phase 1b data in a selected population, and OS was exploratory. Excluding patients with active liver metastases enriches for a group with better prognosis independent of treatment, so comparison with historical later-line outcomes is unreliable, as the coverage itself notes. After an end-of-phase 2 meeting in 2024, the FDA advised against seeking accelerated approval on the basis of response rate, citing uncertainty about whether response predicts survival benefit. The results have not yet been peer reviewed.
The 3-year survival plateau and the proportion of patients alive and off treatment suggest that a subset of patients with MSS colorectal cancer without active liver metastases may derive durable benefit from dual checkpoint blockade, which would be a meaningful advance for a disease long considered immunologically cold. Definitive evidence will come from the randomized phase 3 BATTMAN trial (NCT07152821), comparing botensilimab plus balstilimab and best supportive care with best supportive care alone in refractory MSS/pMMR disease. Understanding and overcoming liver-associated immune resistance remains a central challenge.
Source: C-800-01 phase 1b study, presented at the ESMO Gastrointestinal Cancers Congress 2026, Munich, July 2026; not yet published in a peer-reviewed journal. Conference coverage (Targeted Oncology)
7. Treatment delay thresholds and metastasis risk in nonmetastatic colorectal cancer (JAMA Network Open)
Chi M. Nguyen (Department of Biostatistics & Health Data Science, Indiana University School of Medicine), T. C. Skaar, T. F. Imperiale, et al.; corresponding author Chi M. Nguyen
Key points
- Retrospective cohort study of 11,927 insured US adults aged 40 or older with newly diagnosed nonmetastatic colorectal cancer (2017-2021) who underwent curative-intent surgery within 1 year, using Optum's Clinformatics Data Mart claims.
- Over 3 years, 1,438 patients (12.1%) developed metastatic disease; time to treatment initiation (TTI) was analyzed within four treatment pathways using competing-risk methods.
- Surgery followed by adjuvant therapy: TTI of 4-46 days (sHR 1.27; 95% CI 1.04-1.55) and 47 days or more (sHR 1.55; 95% CI 1.08-2.23) were associated with higher metastasis risk versus 0-3 days.
- Neoadjuvant therapy followed by surgery: TTI of 68 days or more was associated with higher risk (sHR 2.66; 95% CI 1.02-6.94).
- Surgery alone: only delays of 223 days or more suggested higher risk, not statistically significant; trimodality therapy: no significant association.
Key limitations
- Claims data lack detailed tumor characteristics such as stage detail and pathology.
- Treatment timing derived from claims may misclassify time to treatment initiation; residual confounding is possible.
- Commercially insured and Medicare Advantage population; findings may not generalize to uninsured or underserved patients.
Future directions
- Validation in registry cohorts with full clinical and pathologic detail.
- Development of pathway-specific quality benchmarks for timeliness of colorectal cancer care.
- Interventions to reduce avoidable delays, particularly for patients treated with surgery and adjuvant chemotherapy.
Timeliness of cancer treatment is a widely used quality measure, and delays between diagnosis and the start of treatment are commonly thought to worsen outcomes. For colorectal cancer, however, the effect of delay is difficult to study because the appropriate interval differs greatly by treatment pathway: a patient going straight to surgery, one receiving neoadjuvant chemoradiotherapy, and one requiring staging and multidisciplinary planning follow very different timelines, and a single universal threshold may misclassify appropriate sequencing as delay.
Studies that pool patients across treatment strategies, or rely on overall survival, which is influenced by many factors other than timing, may obscure the effect of delay. Nguyen and colleagues sought to define delay thresholds specific to each treatment pathway and to examine an outcome more directly tied to disease progression: development of metastatic disease within 3 years of diagnosis. Metastatic progression is also the event that prompt treatment is most plausibly intended to prevent.
The investigators used deidentified administrative claims from Optum's Clinformatics Data Mart to identify 11,927 adults aged 40 or older diagnosed with nonmetastatic colorectal cancer between 2017 and 2021 who underwent curative-intent surgery within 1 year of diagnosis. Patients were assigned to one of four pathways: surgery with or without radiation; neoadjuvant therapy followed by surgery; surgery followed by adjuvant therapy; and trimodality therapy. The exposure was time to treatment initiation (TTI), the number of days from diagnosis to the first cancer-directed treatment, and the primary outcome was metastatic disease within 3 years, analysed with competing-risk methods.
Over 3 years of follow-up, 1,438 patients (12.1%) developed metastatic colorectal cancer. Thresholds and associations differed by pathway. For patients treated with surgery followed by adjuvant therapy, starting treatment 4 to 46 days after diagnosis was associated with a 27% higher risk of metastasis than starting within 3 days (subdistribution HR 1.27; 95% CI 1.04-1.55), and a delay of 47 days or more with a 55% higher risk (sHR 1.55; 95% CI 1.08-2.23). For patients receiving neoadjuvant therapy before surgery, starting after 68 days or more was associated with a substantially higher risk (sHR 2.66; 95% CI 1.02-6.94).
For patients treated with surgery alone, only very long delays of 223 days or more suggested higher risk, and this association did not reach statistical significance. Among patients receiving trimodality therapy, no significant association between TTI and metastasis was found. The authors emphasize that longer intervals in multimodal pathways may reflect appropriate sequencing and planning rather than harmful delay, and that pathway-specific benchmarks could identify avoidable delays without penalizing appropriate timelines.
The study has the limitations inherent in claims data. Detailed tumor characteristics are not captured, so residual confounding by stage, tumor biology or comorbidity is possible; for example, patients treated very quickly may differ systematically from those treated later. Treatment dates inferred from claims may misclassify TTI, and the very short reference interval (0-3 days) for the adjuvant pathway may itself select an unusual group. The cohort was drawn from commercially insured and Medicare Advantage beneficiaries, which may limit generalizability to underserved populations, and the confidence interval for the neoadjuvant estimate is wide.
The study suggests that the timeliness of colorectal cancer care should be measured against pathway-specific thresholds rather than a single universal standard. The authors conclude that their findings "underscore the importance of coordinated, timely care" and that "ensuring prompt access to CRC treatment remains an evidence-based priority." Validation in clinical registries with full pathologic data, and testing of interventions that shorten avoidable delays, especially for patients on the surgery-then-adjuvant pathway, are logical next steps.
Source: Nguyen et al., "Colorectal Cancer Treatment Delay Thresholds and Metastasis Risk," JAMA Network Open, July 14, 2026. Read the paper