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. A single-nucleus transcriptomic atlas of early-onset colorectal cancer (Nature Medicine)
Caroline Porter (Klarman Cell Observatory, Broad Institute), et al.; co-senior authors Ana C. Anderson, Kimmie Ng, and Ramnik J. Xavier
Key points
- Single-nucleus RNA sequencing (snRNA-seq) of 189 primary colorectal adenocarcinomas from patients aged 29 to more than 90 years (54 diagnosed before age 50; 135 at age 50 or older), comprising more than 850,000 nuclei.
- In microsatellite-stable (MSS) tumors, early-onset disease showed a higher stromal-to-malignant epithelial cell ratio, a relationship that inverted with advancing age.
- Malignant cells from younger patients showed increased long noncoding RNA (lncRNA) expression and enrichment of epithelial-to-mesenchymal transition (EMT) programs, most evident in left-sided MSS tumors.
- Findings were corroborated in bulk transcriptomic data from 564 tumors and digital histopathology from 712 tumors.
- The design is observational and cross-sectional; no causal link to clinical aggressiveness was established.
Key limitations
- Cross-sectional, observational design; associations cannot establish that stromal enrichment or EMT activity causes clinical aggressiveness.
- Some analyses relied on smaller tumor subsets, and residual confounding could not be fully excluded.
- Few immune cells were captured, precluding conclusions about age-related differences in antitumor immunity.
Future directions
- Functional perturbation of the implicated lncRNAs, already under way in vitro.
- Evaluation of stromal content as a predictive biomarker and of stroma-directed therapy in early-onset disease.
- Integration with exposure and birth-cohort data to link tumor biology to the causes of rising incidence.
The incidence of colorectal cancer among adults younger than 50 has risen steadily since the mid-1990s, even as incidence and mortality have fallen in older, screening-eligible populations. Colorectal cancer is now the leading cause of cancer death among Americans under 50, and rising incidence has been documented even in adolescents. Early-onset colorectal cancer (EOCRC) is clinically characterized by frequent presentation at advanced stage, a predilection for the distal colon and rectum, and an apparently aggressive course, yet whether it is a biologically distinct entity or simply conventional colorectal cancer arising earlier in life has remained unresolved.
Previous molecular comparisons of early- and late-onset disease have relied largely on bulk genomic and transcriptomic profiling, which averages signals across heterogeneous cell populations and cannot readily separate tumor-intrinsic programs from differences in the composition of the tumor microenvironment. Many such studies have also dichotomized patients at age 50, an administrative threshold that may obscure gradual biological change across the lifespan. Single-cell approaches offer the resolution to address both limitations, but have until now been applied to cohorts too small to model age as a continuous variable.
Porter and colleagues, working across the Broad Institute's Klarman Cell Observatory, Dana-Farber Cancer Institute, Brigham and Women's Hospital, and Massachusetts General Hospital, performed snRNA-seq on archived frozen surgical specimens from Dana-Farber's Gastrointestinal Biobank. Using single-nucleus rather than single-cell sequencing made it possible to profile banked frozen tissue at scale. Age was modeled across its full range rather than as a binary variable, and analyses were adjusted for tumor location, stage, sex, and prior treatment where appropriate. Key observations were validated in an independent bulk transcriptomic dataset of 564 tumors and in image-based analysis of stained tissue from 712 tumors.
The central compositional finding was that MSS tumors from younger patients contained a higher proportion of stromal cells relative to malignant epithelial cells, a ratio that declined and ultimately reversed in older patients. Stromal populations can produce immunosuppressive mediators and may impede drug delivery, and high stromal content has been associated with chemoresistance in other solid tumors, offering a plausible though untested link to the poorer treatment responses reported in EOCRC. Within the malignant compartment, younger patients' tumors showed greater expression of lncRNAs, many of uncharacterized function, and enrichment of pathways associated with EMT, a program implicated in invasion and metastatic dissemination. These signals were most pronounced in left-sided MSS tumors, the subgroup that accounts for much of the clinical burden of EOCRC.
Several features strengthen these observations: the size of the cohort relative to prior single-cell studies, the treatment of age as a continuum, and orthogonal validation across sequencing and imaging modalities. The age-associated differences were also observed specifically within MSS disease, which reduces the risk that they simply reflect the different representation of mismatch repair-deficient tumors across age groups.
The authors acknowledge important limitations. Some analyses relied on smaller subsets of tumors, residual confounding could not be fully excluded, and the relatively small number of immune cells captured precludes firm conclusions about age-related differences in antitumor immunity. Because the design is cross-sectional, the work cannot establish whether stromal enrichment or EMT activity drives clinical aggressiveness, nor does it address the environmental or host exposures thought to underlie the rising incidence of EOCRC.
The atlas nonetheless provides the most detailed cellular map of EOCRC to date and nominates specific, testable hypotheses. The investigators are already conducting functional perturbation of the implicated lncRNAs in vitro, and the stromal findings invite study of whether stroma-directed strategies, or stromal content as a biomarker, could inform treatment in younger patients. Co-senior author Ana Anderson characterized the work as a potential turning point for a field that has had "more questions than answers."
Source: Porter et al., "Insights into early-onset colorectal cancer from a single-nucleus RNA-seq atlas of colorectal cancer across the age continuum," Nature Medicine, October 29, 2026. Read the paper
2. Short-course radiotherapy followed by camrelizumab and CAPOX in locally advanced rectal cancer (UNION, The Lancet Oncology)
Presenting author T. Zhang (Wuhan, China), on behalf of the UNION investigators
Key points
- Randomized, multicenter phase 3 trial in China (NCT04928807) of 231 patients with locally advanced rectal adenocarcinoma (cT3-4 or node-positive, within 10 cm of the anal verge).
- Experimental arm: short-course radiotherapy (25 Gy in 5 fractions), then neoadjuvant CAPOX plus the anti-PD-1 antibody camrelizumab; control arm: long-course chemoradiotherapy (50.4 Gy in 28 fractions with capecitabine), then neoadjuvant CAPOX.
- Pathologic complete response (pCR), the primary endpoint, was reported previously: 39.8% vs 15.3% (odds ratio 3.7; 95% CI 2.0-6.9; P<0.001).
- 3-year event-free survival (EFS): 91.3% vs 77.4% (HR 0.33; 95% CI 0.15-0.71; P=0.0029).
- Grade 3 or higher treatment-related adverse events: 51.3% vs 48.2%.
Key limitations
- Two variables changed at once (radiotherapy fractionation and camrelizumab), so the contribution of each cannot be isolated.
- EFS was a secondary, investigator-assessed endpoint; overall survival is immature and pMMR/MSS-specific EFS was not reported.
- Single-country trial using a PD-1 antibody not widely available outside China.
Future directions
- Component-isolating or factorial trials to separate the effects of short-course radiotherapy and PD-1 blockade.
- Biomarker studies to identify which pMMR/MSS patients benefit.
- Replication in Western populations with globally available PD-1 inhibitors, and mature data on local control, late toxicity, and organ preservation.
The management of locally advanced rectal cancer has shifted toward total neoadjuvant therapy, in which systemic chemotherapy and radiotherapy are delivered before total mesorectal excision with the aim of improving systemic control, increasing complete response rates, and creating opportunities for organ preservation. Immune checkpoint blockade has produced remarkable results in mismatch repair-deficient (dMMR) rectal cancer, but the large majority of rectal tumors are mismatch repair-proficient or microsatellite stable (pMMR/MSS), a phenotype largely refractory to PD-1 blockade as monotherapy.
A biological rationale has emerged for combining radiotherapy with immunotherapy in pMMR disease. Radiation can induce immunogenic cell death, increase antigen presentation, and remodel the tumor microenvironment, and preclinical and early clinical data suggest that hypofractionated short-course regimens may be more immunologically favorable than conventional long-course chemoradiotherapy, which delivers prolonged exposure to circulating and tumor-infiltrating lymphocytes. Earlier phase 2 studies had reported encouraging response rates with short-course radiotherapy followed by chemotherapy and PD-1 inhibition, setting the stage for a definitive randomized comparison.
UNION enrolled 231 patients between July 2021 and March 2023 and randomized them 1:1. The experimental arm received short-course radiotherapy (25 Gy in 5 fractions), followed by two cycles of neoadjuvant CAPOX with camrelizumab 200 mg and six cycles of adjuvant CAPOX after surgery. The control arm received long-course chemoradiotherapy (50.4 Gy in 28 fractions with concurrent capecitabine), followed by two neoadjuvant and six adjuvant cycles of CAPOX without immunotherapy. The primary endpoint was pCR assessed by blinded independent review; EFS was a key secondary endpoint.
Results for the primary endpoint, reported earlier in Annals of Oncology, showed a pCR rate of 39.8% with the experimental regimen versus 15.3% with standard chemoradiotherapy (OR 3.7; 95% CI 2.0-6.9; P<0.001). The present report, at a data cutoff of October 1, 2025 and a median follow-up of approximately 36 months in both arms, demonstrates that this pathologic benefit translated into improved longer-term disease control: 3-year EFS was 91.3% (95% CI 83.9-95.4) versus 77.4% (95% CI 67.9-84.4), an absolute difference of 13.9 percentage points and a hazard ratio of 0.33 (95% CI 0.15-0.71; P=0.0029).
Toxicity was comparable between arms, with grade 3 or higher treatment-related adverse events in 51.3% (59/115) of the experimental arm and 48.2% (54/112) of the control arm. Overall survival data were not reported at this analysis. The magnitude of EFS benefit is notable for a predominantly pMMR population, in which checkpoint inhibition has historically conferred little activity.
Interpretation requires several caveats. Because the experimental arm differed from the control in both radiotherapy fractionation and the addition of camrelizumab, the design cannot isolate the contribution of either component; the observed benefit may reflect immunotherapy, the change in fractionation, or their interaction. EFS was a secondary, investigator-assessed endpoint, results were not reported separately for the pMMR/MSS subgroup, overall survival remains immature, and the trial was conducted exclusively in China with a PD-1 antibody not widely available elsewhere. Commentators have also noted that mature data on local control, late toxicity, and organ preservation are needed before short-course radiotherapy is accepted as the preferred partner for immunotherapy.
If confirmed, UNION would represent one of the first phase 3 demonstrations that adding PD-1 blockade to a total neoadjuvant strategy improves outcomes in largely pMMR rectal cancer. The investigators describe the regimen as a potential new standard-of-care option. Factorial or component-isolating trials, biomarker work to identify which pMMR patients benefit, and replication in Western populations with globally available PD-1 inhibitors will be essential next steps.
Source: UNION trial, The Lancet Oncology, published online October 25, 2026. Read the paper · PubMed
3. Biweekly versus standard trifluridine/tipiracil plus bevacizumab in refractory metastatic colorectal cancer (PRABITAS, The Lancet)
Presenting author Hiroya Taniguchi (Department of Clinical Oncology, Aichi Cancer Center Hospital, Nagoya, Japan), on behalf of the PRABITAS investigators
Key points
- Pragmatic, randomized, open-label phase 3 noninferiority trial (jRCTs041230120) at 178 hospitals in Japan; 972 patients randomized, 970 in the intention-to-treat analysis.
- A 14-day (biweekly) schedule of trifluridine/tipiracil (FTD/TPI) plus bevacizumab was compared with the conventional 28-day schedule; the noninferiority margin for overall survival (OS) was a hazard ratio of 1.25.
- Median OS was 11.3 vs 10.3 months (HR 0.93; 95% CI 0.80-1.08), meeting noninferiority but not superiority (P=0.35).
- Grade 3 or higher neutropenia: 13% vs 39%; febrile neutropenia: 1% vs 5%; grade 2 or higher anorexia: 15% vs 26%.
- Two treatment-related deaths (gastrointestinal hemorrhage, perforation) occurred in the biweekly arm, both attributed to bevacizumab.
Key limitations
- Conducted exclusively in Japan; generalizability to other populations is uncertain.
- Patient-reported outcomes and quality of life were not formally assessed.
- Open-label design; two bevacizumab-related deaths in the biweekly arm warrant continued safety monitoring.
Future directions
- Confirmation of tolerability and efficacy in non-Asian populations.
- Prospective collection of patient-reported outcomes to quantify the clinical value of reduced toxicity.
- Incorporation of the biweekly schedule into guidelines and its evaluation with newer later-line agents.
Trifluridine/tipiracil (FTD/TPI) is a standard later-line therapy for metastatic colorectal cancer after progression on fluoropyrimidine-, oxaliplatin-, and irinotecan-based regimens, and its combination with bevacizumab has become a preferred option in this setting. The conventional schedule, oral dosing twice daily on days 1-5 and 8-12 of each 28-day cycle, is limited principally by myelosuppression. Neutropenia frequently necessitates dose delays and reductions in a population for whom preserving function and quality of life is a central therapeutic goal.
Alternative biweekly schedules, in which FTD/TPI is given for five consecutive days every 14 days, were developed to reduce cumulative marrow toxicity while maintaining dose intensity. Earlier single-arm and randomized phase 2 data suggested that biweekly administration, alone or with bevacizumab, reduced hematologic toxicity with apparently preserved efficacy, and the schedule had entered clinical use in parts of Japan. A definitive randomized assessment of its effect on survival had, however, been lacking.
PRABITAS was designed as a pragmatic, open-label, noninferiority trial embedded in routine practice across 178 Japanese hospitals. Patients with refractory metastatic colorectal cancer were randomized 1:1 to FTD/TPI plus bevacizumab on the biweekly schedule or the conventional 28-day schedule. The primary endpoint was OS, with noninferiority defined by an upper confidence bound for the hazard ratio below 1.25. A total of 972 patients were randomized and 970 were included in the intention-to-treat analysis; median follow-up was 17.9 months.
The biweekly schedule met the prespecified noninferiority criterion. Median OS was 11.3 months with biweekly dosing versus 10.3 months with conventional dosing (HR 0.93; 95% CI 0.80-1.08), and superiority was not demonstrated (P=0.35). Median progression-free survival was identical at 3.9 months in both arms (HR 0.93; 95% CI 0.81-1.06).
The principal benefit was in tolerability. Grade 3 or higher neutropenia fell from 39% with the conventional schedule to 13% with biweekly dosing, febrile neutropenia from 5% to 1%, and grade 2 or higher anorexia from 26% to 15%. Two treatment-related deaths occurred in the biweekly arm, one from gastrointestinal hemorrhage and one from gastrointestinal perforation, both attributed to bevacizumab; none occurred in the conventional arm.
The trial has several limitations. It enrolled patients exclusively in Japan, where FTD/TPI pharmacokinetics, supportive care practices, and patterns of prior therapy may differ from those elsewhere. Patient-reported outcomes and quality of life were not formally assessed, so the clinical meaning of the reduced toxicity for patients cannot be quantified directly. The open-label design is appropriate for an OS endpoint but may have influenced the attribution and reporting of adverse events, and the two bevacizumab-related deaths in the biweekly arm, although probably unrelated to the schedule, warrant continued vigilance.
PRABITAS provides high-level evidence that a simpler biweekly FTD/TPI schedule can be used without compromising survival while substantially reducing hematologic toxicity. Because the trial was large, pragmatic, and run in routine practice, its results are likely to apply directly to clinical care, and the biweekly regimen is a reasonable default for patients at risk of myelosuppression. Confirmation of tolerability and patient-reported benefit in non-Asian populations would strengthen its adoption internationally.
Source: PRABITAS trial, The Lancet, published online October 24, 2026. Read the paper
4. Adjuvant mFOLFIRINOX versus mFOLFOX6 in high-risk stage III colon cancer (IROCAS, Annals of Oncology)
J. Bennouna (Suresnes, France), T. André, E. Martin, C. Cremolini, S. Gill, et al. (33 authors), C. O’Callaghan, J. Taieb
Key points
- International, randomized phase 3 trial (NCT02967289; PRODIGE 52) conducted by PRODIGE, the Canadian Cancer Trials Group, and GONO; 795 patients with R0-resected pT4 and/or pN2 colon adenocarcinoma.
- Twelve cycles of mFOLFIRINOX (n=398) versus mFOLFOX6 (n=397); primary endpoint 3-year disease-free survival (DFS).
- 3-year DFS: 65.6% vs 66.3% (HR 1.07; 95% CI 0.84-1.36; P=0.61).
- 3-year overall survival: 91.2% vs 86.7% (HR 0.85; 95% CI 0.61-1.21; P=0.37), immature and not significant.
- Grade 3 or higher adverse events were similar overall (62.2% vs 62.1%), but grade 3 or higher diarrhea (15.6% vs 3.8%), fatigue (7.3% vs 1.8%), and G-CSF use (37.8% vs 19.1%) were higher with mFOLFIRINOX.
Key limitations
- Primary endpoint not met; the numerical overall survival difference is immature and not statistically significant.
- No molecular selection (e.g., mismatch repair status or ctDNA), which may have diluted effects in sensitive subgroups.
- Substantially greater gastrointestinal toxicity and G-CSF use with the triplet.
Future directions
- Mature overall survival analysis.
- ctDNA-guided escalation strategies to identify patients who need more intensive adjuvant therapy.
- Neoadjuvant chemotherapy and immunotherapy (for dMMR tumors) as alternatives to adjuvant intensification.
Adjuvant fluoropyrimidine and oxaliplatin chemotherapy (FOLFOX or CAPOX) is the standard of care after curative resection of stage III colon cancer. Outcomes are heterogeneous, however, and patients with pT4 and/or pN2 disease constitute a high-risk subgroup in whom roughly one-third experience recurrence or death within three years despite standard therapy. Intensification of adjuvant treatment for this population has therefore been a long-standing research priority.
The rationale for adding irinotecan derived from the activity of the FOLFIRINOX and FOLFOXIRI triplets in metastatic colorectal cancer and of mFOLFIRINOX in the adjuvant treatment of pancreatic cancer. Historically, however, irinotecan-containing regimens failed to improve outcomes in the adjuvant setting for colon cancer when combined with fluoropyrimidines in earlier randomized trials, raising the question of whether a modern triplet incorporating oxaliplatin would perform differently in a molecularly unselected high-risk population.
IROCAS randomized 795 patients between May 2017 and June 2023 to 12 biweekly cycles of mFOLFIRINOX (oxaliplatin 85 mg/m², irinotecan 180 mg/m², leucovorin 400 mg/m², and infusional 5-FU 2,400 mg/m² over 46 hours) or mFOLFOX6. Eligible patients had R0-resected colon adenocarcinoma classified as pT4 and/or pN2; 52.8% had pT4 tumors, 72.2% had pN2 disease, and 25.3% had both. The median age was 59 years and median follow-up was 55 months.
The trial did not meet its primary endpoint. Three-year DFS was 65.6% with mFOLFIRINOX and 66.3% with mFOLFOX6 (HR 1.07; 95% CI 0.84-1.36; P=0.61), with 146 and 138 DFS events, respectively. No statistically significant treatment-by-subgroup interaction was identified across prespecified subgroups. Three-year overall survival was numerically higher with mFOLFIRINOX (91.2% vs 86.7%), but the difference was not significant (HR 0.85; 95% CI 0.61-1.21; P=0.37) and the data remain immature.
The overall rate of grade 3 or higher adverse events was virtually identical between arms (62.2% vs 62.1%), but the toxicity profile differed. mFOLFIRINOX was associated with substantially more grade 3 or higher diarrhea (15.6% vs 3.8%) and fatigue (7.3% vs 1.8%) and with greater use of granulocyte colony-stimulating factor (37.8% vs 19.1%), whereas grade 3 or higher neutropenia was slightly less frequent (18.0% vs 23.9%), plausibly reflecting more frequent G-CSF prophylaxis.
The principal limitation in interpreting IROCAS is the immaturity of the overall survival data; although the numerical difference favored mFOLFIRINOX, the confidence interval is wide and compatible with no effect, and a post hoc emphasis on this secondary endpoint would be inappropriate. The trial also did not stratify by molecular features such as mismatch repair status or circulating tumor DNA, which may define subpopulations with differential sensitivity to intensified chemotherapy.
IROCAS adds to a consistent body of evidence that irinotecan does not improve outcomes in the adjuvant treatment of colon cancer, and it argues against chemotherapy intensification as a strategy for unselected high-risk stage III disease. As one commentator summarized, adding irinotecan "did nothing." Future progress in this population is more likely to come from biologically selected approaches, such as ctDNA-guided escalation, neoadjuvant chemotherapy, and immunotherapy for dMMR tumors.
Source: IROCAS trial, Annals of Oncology, October 2026. Read the paper
5. Circulating tumor DNA methylation-guided surveillance after curative resection (FIND, Journal of Clinical Oncology)
Mo, et al.; corresponding author Junjie Peng (Fudan University Shanghai Cancer Center)
Key points
- Investigator-initiated, multicenter, randomized phase 3 trial at six academic centers in China; 728 patients with stage I-III colorectal cancer randomized after R0 resection, 584 in the modified intention-to-treat population (289 ctDNA-guided, 295 control).
- The intervention was plasma ctDNA methylation testing every three months for two years, with immediate CT imaging after a positive result; the control was guideline-based CT and CEA surveillance.
- Curative-intent metastasis-directed therapy among patients with recurrence: 48.1% vs 23.6% (relative risk 2.03).
- Recurrence rates were similar (18.0% vs 18.6%), with a median lead time of 3.9 months for ctDNA-guided detection.
- Sensitivity for recurrence: 84.1% for ctDNA versus 35.7% for CEA; overall survival data are immature.
Key limitations
- Overall survival is immature; earlier curative-intent treatment may not translate into longer survival, and lead-time bias must be considered.
- A single methylation-based assay was tested; results may not generalize to other ctDNA platforms.
- Many randomized patients were excluded from the modified intention-to-treat analysis; cost-effectiveness and patient-reported outcomes are pending.
Future directions
- Mature overall survival data to determine whether ctDNA-guided surveillance improves survival.
- Head-to-head comparison of ctDNA assay platforms and testing intervals.
- Health-economic evaluation and assessment of the psychological burden of frequent testing and false-positive results.
Postoperative surveillance after curative resection of colorectal cancer aims to detect recurrence while it remains amenable to curative-intent treatment, most commonly resection or ablation of oligometastatic disease in the liver or lungs. Current guideline-based surveillance combines serial carcinoembryonic antigen (CEA) measurement with cross-sectional imaging, but CEA has limited sensitivity, and several large randomized trials of intensive versus less intensive follow-up have failed to show a survival advantage from more frequent conventional testing.
Circulating tumor DNA has emerged as a highly specific marker of minimal residual disease after surgery, and observational cohorts have shown that postoperative ctDNA positivity strongly predicts recurrence, often months before radiologic detection. Randomized evidence has so far focused largely on ctDNA-guided adjuvant therapy decisions; whether ctDNA-guided surveillance changes the management of recurrence had not been established in a randomized setting.
FIND randomized patients with stage I-III colorectal cancer who had undergone R0 resection and had no radiologically detectable distant metastases. In the intervention arm, plasma ctDNA methylation testing was performed preoperatively, within one month after surgery, and every three months for two years, and any positive result prompted immediate CT of the chest, abdomen, and pelvis. The control arm received guideline-based CT and CEA surveillance. The primary endpoint was the proportion of patients with radiologically confirmed recurrence who received curative-intent metastasis-directed therapy. At the February 2026 data cutoff, median follow-up was 23.3 months.
The primary endpoint was met. Among patients who experienced recurrence, 48.1% in the ctDNA-guided arm received curative-intent metastasis-directed therapy, compared with 23.6% in the control arm (relative risk 2.03; absolute difference 24.4 percentage points). Recurrence rates did not differ (18.0% [52/289] vs 18.6% [55/295]), indicating that the benefit reflected earlier detection rather than a change in underlying recurrence risk. Median time to clinical recurrence was 9.5 months versus 13.4 months, a median lead time of 3.9 months.
Secondary analyses were consistent with this mechanism. Among recurrences confined to the liver and/or lungs, curative-intent treatment was delivered in 42.3% of ctDNA-guided patients versus 18.2% of controls. For detection of recurrence, the ctDNA assay achieved a sensitivity of 84.1% and specificity of 91.4%, compared with 35.7% and 94.0% for CEA. Post-recurrence progression-free survival also favored the ctDNA arm (HR 0.51).
Several limitations temper interpretation. Overall survival, the endpoint of greatest clinical relevance, remains immature, and earlier curative-intent treatment does not necessarily translate into longer survival; lead-time effects must also be considered. The study evaluated a single methylation-based assay, so its findings cannot be assumed to generalize to tumor-informed mutation-based platforms with different performance characteristics. Health-economic analyses and long-term patient-reported outcomes, including the psychological consequences of frequent testing and false-positive results, are pending. A substantial proportion of randomized patients were excluded from the modified intention-to-treat population.
FIND provides the first randomized evidence that ctDNA-guided surveillance can materially increase the proportion of recurrences treated with curative intent, a necessary, though not sufficient, condition for improving survival. If mature overall survival data confirm a benefit, ctDNA-based monitoring may be incorporated into surveillance guidelines; in the interim, the trial supports continued investigation of ctDNA as a complement to imaging. This article was first published online in July 2026 and appears in the journal's current issue.
Source: FIND trial, Journal of Clinical Oncology, first published online July 29, 2026. Read the paper · PubMed
6. m-FOLFOXIRI plus cetuximab versus bevacizumab in RAS wild-type metastatic colorectal cancer: final analysis (DEEPER/JACCRO CC-13, Journal of Clinical Oncology)
Akihito Tsuji, et al., for the JACCRO CC-13 (DEEPER) investigators
Key points
- Randomized phase 2 trial in Japan of 359 patients with RAS wild-type metastatic colorectal cancer (per-protocol set, 321); primary endpoint depth of response (DpR), reported previously.
- Final analysis with follow-up extended to August 2024 because of limited overall survival (OS) events at the primary analysis.
- Left-sided RAS wild-type tumors: PFS 13.9 vs 12.1 months (HR 0.81; 95% CI 0.63-1.05); OS 45.3 vs 41.9 months (HR 0.85; 95% CI 0.64-1.12).
- Left-sided RAS/BRAF wild-type tumors (exploratory, n=178): PFS 14.8 vs 11.9 months (HR 0.71; 95% CI 0.52-0.97); OS 50.2 vs 40.2 months (HR 0.74; 95% CI 0.53-1.05).
- The advantage concentrated in patients with extrahepatic metastases (OS HR 0.60; 95% CI 0.40-0.90) and was absent in liver-limited disease.
Key limitations
- Phase 2 trial powered for depth of response, not survival; survival analyses are secondary.
- The most pronounced effects come from exploratory subgroups with wide confidence intervals; the main left-sided comparisons were not significant.
- BRAF status was missing for over a quarter of patients, introducing potential selection bias.
Future directions
- Prospective validation of anti-EGFR triplet therapy in left-sided RAS/BRAF wild-type disease.
- Investigation of metastatic distribution (extrahepatic vs liver-limited) as a factor in choosing the biologic partner.
- Integration of liquid biopsy-based molecular selection into first-line treatment choice.
In first-line treatment of metastatic colorectal cancer, the choice of biologic partner for cytotoxic chemotherapy is guided by primary tumor sidedness and molecular status. Retrospective analyses of large randomized trials established that anti-EGFR antibodies confer greater benefit than bevacizumab in left-sided RAS wild-type tumors, a finding later confirmed prospectively with doublet chemotherapy. For fit patients, the triplet FOLFOXIRI achieves higher response rates than doublets and has been combined most extensively with bevacizumab.
Whether an anti-EGFR antibody is the preferred partner for triplet chemotherapy has remained uncertain, in part because of concerns about overlapping gastrointestinal and cutaneous toxicity and because depth and speed of response, rather than response rate alone, have been proposed as more informative surrogates of benefit with intensive regimens. DEEPER (JACCRO CC-13) was designed to compare cetuximab and bevacizumab added to modified FOLFOXIRI, using depth of response as its primary endpoint.
The trial randomized 359 patients with RAS wild-type metastatic colorectal cancer to m-FOLFOXIRI (irinotecan 150 mg/m², oxaliplatin 85 mg/m², and 5-FU 2,400 mg/m²) plus either cetuximab or bevacizumab. After the primary DpR analysis, observation was extended because relatively few OS events had accrued; the final analysis used an August 2024 cutoff, a prespecified per-protocol set of 321 patients, and two-sided log-rank testing. BRAF status was available for 234 of 321 patients (73%).
Among patients with left-sided RAS wild-type tumors, cetuximab was associated with numerically longer progression-free survival (median 13.9 vs 12.1 months; HR 0.81; 95% CI 0.63-1.05) and overall survival (median 45.3 vs 41.9 months; HR 0.85; 95% CI 0.64-1.12), but neither comparison reached statistical significance. In the exploratory subgroup of 178 patients with left-sided RAS/BRAF wild-type tumors, the difference widened: median PFS was 14.8 versus 11.9 months (HR 0.71; 95% CI 0.52-0.97) and median OS 50.2 versus 40.2 months (HR 0.74; 95% CI 0.53-1.05).
Stratification by metastatic distribution revealed marked heterogeneity. In left-sided RAS/BRAF wild-type patients with extrahepatic metastases, cetuximab was associated with longer PFS (15.1 vs 11.4 months; HR 0.66; 95% CI 0.46-0.95) and OS (50.2 vs 38.6 months; HR 0.60; 95% CI 0.40-0.90). In patients with liver-limited disease, by contrast, outcomes were similar or numerically favored bevacizumab (OS 52.2 vs 49.9 months; HR 1.17; 95% CI 0.61-2.24), possibly reflecting the role of secondary resection in that population.
These findings must be interpreted cautiously. DEEPER is a phase 2 trial powered for depth of response rather than survival; the survival analyses are secondary, the most pronounced effects arise in exploratory subgroups with wide confidence intervals, and the primary left-sided comparisons did not achieve significance. Missing BRAF data for over a quarter of patients introduces potential selection bias, and the analysis by metastatic site was not designed to support definitive conclusions.
Nevertheless, the final DEEPER results are directionally concordant with the broader evidence base favoring anti-EGFR therapy in left-sided RAS/BRAF wild-type disease, and they extend that principle to triplet chemotherapy. They suggest that m-FOLFOXIRI plus cetuximab is a reasonable option for fit patients with left-sided, RAS/BRAF wild-type tumors, particularly those with extrahepatic disease, while leaving room for bevacizumab-based triplets in liver-limited disease. These final results were first presented at the ASCO Gastrointestinal Cancers Symposium in January 2025.
Source: DEEPER trial (JACCRO CC-13), final analysis, Journal of Clinical Oncology, 2026. PubMed
7. Dostarlimab monotherapy in dMMR/MSI-H locally advanced rectal cancer (AZUR-1, ESMO Congress 2026)
Andrea Cercek (Memorial Sloan Kettering Cancer Center), J.-B. Bachet, G. Folprecht, et al.
Key points
- Global, single-arm phase 2 trial (NCT05723562) of 154 patients with previously untreated stage II or III mismatch repair-deficient or microsatellite instability-high (dMMR/MSI-H) rectal cancer.
- Dostarlimab 500 mg every three weeks for nine cycles, followed by endoscopic and radiologic restaging; patients with clinical complete response (cCR) entered nonoperative surveillance.
- Primary endpoint, cCR sustained for at least 12 months (cCR12) by independent central review: 47% (data 75% mature); investigator-assessed cCR12: 68%.
- Among 100 patients with cCR and at least 12 months of follow-up, there were no recurrences, and 98% maintained cCR without further treatment.
- The FDA granted accelerated approval on October 19, 2026. These results were presented at ESMO 2026 (LBA87) and have not yet been published in a peer-reviewed journal.
Key limitations
- Single-arm design with no comparator group.
- Short median follow-up (about two years) relative to the timeline of local regrowth and distant recurrence.
- Marked discordance between central (47%) and investigator (68%) cCR12 assessments; results are not yet peer-reviewed.
Future directions
- Longer follow-up to define durable organ preservation and late recurrence.
- Harmonized criteria for assessing clinical complete response, particularly in mucinous tumors.
- Peer-reviewed publication of the full dataset and universal mismatch repair testing at rectal cancer diagnosis.
Trimodality therapy, comprising chemoradiotherapy or total neoadjuvant therapy followed by total mesorectal excision, achieves high rates of local control in locally advanced rectal cancer but carries substantial long-term morbidity, including bowel, urinary, and sexual dysfunction and, in some patients, permanent colostomy. Organ preservation through watch-and-wait strategies has therefore become an important therapeutic goal. Mismatch repair-deficient tumors, which carry high tumor mutational burdens and abundant neoantigens, are exceptionally sensitive to PD-1 blockade.
A single-center phase 2 study at Memorial Sloan Kettering Cancer Center reported in 2022 that every patient with dMMR locally advanced rectal cancer treated with dostarlimab alone achieved a clinical complete response, allowing omission of chemoradiotherapy and surgery. That result, while striking, came from a small cohort at a single high-volume institution, and AZUR-1 was designed as a larger, global, registrational study to determine whether the approach could be replicated across centers and assessed rigorously by independent review.
AZUR-1 enrolled 154 patients with previously untreated stage II or III dMMR/MSI-H rectal adenocarcinoma without symptomatic obstruction; 86% had stage III disease, 91% had T3 or T4 tumors, 84% had mid or low rectal tumors, and the median age was 61.5 years. Patients received dostarlimab 500 mg intravenously every three weeks for nine cycles, followed by endoscopy and MRI or CT. Those with cCR entered nonoperative management; others could be reassessed or proceed to standard therapy at investigator discretion. The primary endpoint was cCR12 by independent central review, with investigator-assessed cCR12 as a secondary endpoint. Median follow-up was 26.3 months.
By independent central review, cCR12 was 47%, with these data 75% mature; by investigator assessment, cCR12 was 68% (94% mature). Overall, 71% of the full analysis set achieved cCR and entered nonoperative management. Among the 100 patients with cCR and at least 12 months of follow-up, no recurrences were observed, 98% maintained cCR without additional therapy, and the two patients who elected surgery had a pathologic complete response. In an exploratory composite of cCR12 or pathologic complete response after dostarlimab alone, the rate was 76%, and 93% of patients remained on study without evidence of progression.
Dostarlimab was well tolerated. Treatment-related adverse events were grade 1-2 in 64% of patients and grade 3 or higher in 10%; the most common were pruritus, hypothyroidism, and fatigue. Immune-mediated adverse events of grade 3 or higher occurred in 3%, treatment-related discontinuation in 2%, and there were no treatment-related deaths. On the basis of these data, the FDA granted accelerated approval to dostarlimab for adults with stage II or III dMMR/MSI-H locally advanced rectal cancer on October 19, 2026.
Several methodological issues merit attention. The single-arm design precludes direct comparison with standard therapy, and the median follow-up of about two years is short relative to the time course of local regrowth and distant recurrence. The substantial discordance between centrally and locally assessed cCR12 highlights the difficulty of determining complete response from images alone, a difficulty compounded by the 15% of mucinous tumors, which are notoriously challenging to assess on MRI. The investigators attributed the gap to delayed calls by central reviewers lacking clinical context, but this interpretation is their own. Denominators differ across analyses, and the composite endpoint was exploratory.
AZUR-1 nonetheless provides multicenter confirmation that PD-1 blockade alone can achieve durable organ preservation in a large proportion of patients with dMMR/MSI-H rectal cancer, and its regulatory approval is likely to reshape practice for this subgroup. The findings reinforce the importance of universal mismatch repair testing at diagnosis of rectal cancer. Longer follow-up, harmonized response assessment criteria, and peer-reviewed publication of the full dataset will be needed to define long-term outcomes of this nonoperative strategy.
Source: AZUR-1 trial, presented at the ESMO Congress 2026 (abstract LBA87), October 2026. Conference coverage (OncLive)