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. First-line mFOLFOX6, bevacizumab and atezolizumab versus atezolizumab alone in dMMR/MSI-H metastatic colorectal cancer (COMMIT, NRG-GI004/SWOG S1610, Journal of Clinical Oncology)
Caio Max Sao Pedro Rocha Lima (Wake Forest University School of Medicine), G. Yothers, T.J. George, H.S. Hochster, H.K. Sanoff, D.J. Cohen, et al.; senior author Michael J. Overman (The University of Texas MD Anderson Cancer Center)
Key points
- Randomized, open-label NCI cooperative-group trial (NRG-GI004/SWOG S1610; NCT02997228) in previously untreated dMMR/MSI-H metastatic colorectal cancer; 82 patients were randomized between atezolizumab alone and mFOLFOX6 plus bevacizumab plus atezolizumab before accrual was suspended for slow enrollment.
- Primary endpoint, progression-free survival (PFS): median 24.5 months (95% CI 10.1-not estimable) with the triplet vs 5.3 months (95% CI 2.2-18.2) with atezolizumab alone; HR 0.42 (95% CI 0.22-0.80; P=0.0068).
- 12-month PFS 66.7% vs 35.1%; 24-month PFS 53.7% vs 31.6%; objective response rate 86.1% vs 46.0%, with complete responses in 36.1% vs 18.9%.
- Progression as best response occurred in 2.8% vs 32.4%; overall survival did not differ (HR 1.04; 95% CI 0.47-2.28; P=0.90), and the trial was not powered for OS.
- Grade 3 or higher adverse events: 82.9% vs 43.9%, including grade 3-4 neutropenia 26.8% vs 0% and hypertension 19.5% vs 2.4%; four grade 5 events in the combination arm and one (disease progression) with atezolizumab alone.
Key limitations
- Small sample after early closure for slow accrual, with a chemotherapy-plus-bevacizumab arm that was closed after 20 patients following KEYNOTE-177.
- Neither arm reflects today's first-line standards (single-agent pembrolizumab or dual checkpoint blockade), so the comparison is hypothesis-generating; an accompanying editorial argued that very little should change in practice.
- No overall survival difference and substantially more severe toxicity with the triplet, including deaths.
Future directions
- Identify the subgroup with primary resistance to checkpoint inhibition who might benefit from upfront chemotherapy-containing combinations.
- Compare chemo-immunotherapy combinations against current standards such as dual checkpoint blockade.
- Develop biomarkers of early progression on immunotherapy alone to guide treatment intensification.
Mismatch repair-deficient or microsatellite instability-high (dMMR/MSI-H) tumors account for about 5% of metastatic colorectal cancers. These hypermutated tumors are highly sensitive to immune checkpoint inhibition, and single-agent anti-PD-1 therapy has replaced chemotherapy as a first-line standard. Yet a meaningful fraction of patients, about one-third according to the trial's sponsors, show primary resistance or early progression on single-agent immunotherapy, a group in whom an initial treatment choice can be costly. Whether adding chemotherapy and anti-VEGF therapy to checkpoint blockade can prevent this early failure has been an open question.
COMMIT (NRG-GI004/SWOG S1610) was designed before checkpoint inhibitors were established in this setting. It originally randomized patients 1:1:1 to mFOLFOX6 plus bevacizumab, atezolizumab alone, or the combination of mFOLFOX6, bevacizumab and atezolizumab. After KEYNOTE-177 showed the superiority of pembrolizumab over chemotherapy, the chemotherapy-only arm was closed after 20 patients had enrolled, and the trial continued as a two-arm comparison of the triplet against atezolizumab monotherapy with a revised sample size of 100 patients. The rationale was that chemotherapy and bevacizumab might debulk disease rapidly and modulate the tumor microenvironment while immunotherapy provides durable control.
Eligible adults had dMMR (by immunohistochemistry) or MSI-H (by PCR or next-generation sequencing) metastatic colorectal cancer, measurable disease by RECIST v1.1, ECOG performance status 0-2 and no prior chemotherapy for metastatic disease; a 2023 amendment allowed one prior cycle of FOLFOX or CAPOX. Randomization was stratified by BRAF V600E status, metastatic site and prior adjuvant chemotherapy. The primary endpoint was PFS in the intention-to-treat population; secondary endpoints included objective response, overall survival, toxicity, 12-month PFS, disease control and duration of response. Eighty-two patients were randomized between the two atezolizumab-containing arms before accrual was suspended for slow enrollment, and median follow-up was 46 months.
The triplet substantially prolonged PFS: median 24.5 months (95% CI 10.1-not estimable) versus 5.3 months (95% CI 2.2-18.2) with atezolizumab alone, a hazard ratio of 0.42 (95% CI 0.22-0.80; P=0.0068), corresponding to a 58% reduction in the risk of progression or death. PFS rates were 66.7% versus 35.1% at 12 months and 53.7% versus 31.6% at 24 months. A sensitivity analysis adjusting for BRAF mutation, liver metastases, prior chemotherapy, performance status, sidedness, race, sex, age and baseline tumor burden gave consistent results.
Objective responses occurred in 86.1% of patients receiving the triplet versus 46.0% with atezolizumab alone, including complete responses in 36.1% versus 18.9%. The most striking difference was in primary refractoriness: progression as best response was seen in 2.8% versus 32.4%, and the 12-month disease control rate was 64.7% versus 32.4% (P=0.009). Overall survival did not differ (HR 1.04; 95% CI 0.47-2.28; P=0.90), possibly reflecting effective subsequent therapy, although the study was not powered for this endpoint. Grade 3 or higher adverse events of any attribution occurred in 82.9% versus 43.9%, driven by neutropenia (26.8% vs 0%), hypertension (19.5% vs 2.4%) and sepsis (9.7% vs 0%). There were four grade 5 events with the combination, including two sudden deaths and a hepatic hemorrhage after extended hepatectomy, and one with atezolizumab alone, attributed to disease progression. Immune-related adverse events were uncommon and similar between arms.
The trial's main weakness is its size: with only 82 patients in the two-arm comparison after early closure, estimates are imprecise and the upper confidence bound for median PFS could not be estimated. More fundamentally, neither arm represents current first-line care, since single-agent atezolizumab is not an approved standard and dual checkpoint blockade has since become an option. In an accompanying editorial, Lars Henrik Jensen (University Hospital of Southern Denmark, Vejle) described the results as hypothesis-generating rather than practice-changing, answering his own question about what should change in practice with "Very little." The absence of an OS signal and the higher rate of severe and fatal toxicity further temper enthusiasm.
COMMIT nonetheless provides randomized evidence that early progression on checkpoint inhibitor monotherapy, a well-recognized problem in dMMR/MSI-H colorectal cancer, can be largely prevented by adding chemotherapy and bevacizumab. The key clinical question is whether this benefit can be targeted to patients at high risk of primary resistance, such as those with heavy tumor burden or symptomatic disease, without exposing all patients to the added toxicity. Principal investigator Caio Rocha Lima said the combination "may provide a meaningful clinical benefit" for this population, while co-principal investigator Michael Overman emphasized the need for further trials. Prospective comparison against modern immunotherapy standards and validated predictors of early progression will be needed before the triplet can be recommended routinely.
Source: Rocha Lima et al., "COMMIT: A Randomized Study of mFOLFOX6/Bevacizumab/Atezolizumab or Atezolizumab Alone as First-Line Treatment of Deficient DNA Mismatch Repair Metastatic Colorectal Cancer," Journal of Clinical Oncology, published online July 29, 2026. Read the paper · PubMed
2. AI-derived tumor cell density and benefit from irinotecan-intensified chemoradiotherapy in rectal cancer: post hoc analysis of ARISTOTLE (eBioMedicine)
Zhuoyan Shen (Department of Medical Physics and Biomedical Engineering, University College London), D. Brand, M. Simard, et al.; senior author Maria A. Hawkins (University College London)
Key points
- Post hoc analysis of 414 ARISTOTLE patients with digitized pretreatment biopsies (ISRCTN09351447); an AI framework measured tumor cell density (TCD: proportion of tumor cells in tumor plus stroma), split at 0.5 into TCD-high (45%, n=188) and TCD-low (55%, n=226).
- Significant treatment-by-TCD interaction: DFS HR 0.44 (95% CI 0.24-0.82; P=0.009) and OS HR 0.31 (95% CI 0.16-0.64; P=0.001); consistent after adjusting for KRAS and TP53.
- TCD-high: irinotecan improved DFS (HR 0.57; 95% CI 0.36-0.90; P=0.014) and OS (HR 0.50; 95% CI 0.30-0.84; P=0.008); pCR 23% vs 11% (OR 2.46; 95% CI 1.01-5.98; P=0.042, P=0.13 after multiplicity adjustment).
- TCD-low: no benefit, with a trend toward worse OS (DFS HR 1.29; 95% CI 0.85-1.95; OS HR 1.55; 95% CI 0.96-2.51; P=0.07); pCR 14% vs 22%.
- AI tumor region detection: precision 0.873, recall 0.984; agreement with manual TCD, Lin's concordance 0.831 (95% CI 0.760-0.866).
Key limitations
- Post hoc, hypothesis-generating analysis of a negative trial; the 0.5 cutoff was not prespecified in the parent trial.
- The pCR difference in TCD-high tumors did not survive multiple-testing correction, and within-arm comparisons were not all significant.
- No independent external validation cohort; the authors state validation is required before use in treatment selection.
Future directions
- Validation of AI-measured TCD in other neoadjuvant rectal cancer trials, including those using irinotecan-containing total neoadjuvant therapy.
- Prospective, biomarker-stratified trials of intensification in TCD-high tumors.
- Study of the biology linking stromal content to chemoradiotherapy resistance.
Locally advanced rectal cancer is treated with increasingly intensive multimodality regimens, yet trials of treatment intensification have frequently been negative overall while adding toxicity. A likely reason is biological heterogeneity: if only some tumors benefit from a stronger regimen, the effect is diluted in an unselected population, and patients unlikely to benefit are exposed to harm. Tools that predict benefit from routinely collected material, such as diagnostic hematoxylin and eosin (H&E) biopsies, could allow intensification to be targeted.
The ARISTOTLE phase 3 trial, reported separately, found no benefit from adding irinotecan to capecitabine-based chemoradiotherapy in margin-threatened rectal cancer, with more toxicity. The stromal component of colorectal tumors has been linked to poor prognosis and treatment resistance in earlier work. Shen and colleagues at University College London hypothesized that the proportion of tumor cells relative to stroma, measurable with digital pathology, might identify a subgroup in which irinotecan is effective.
The investigators applied an AI framework to digitized pretreatment H&E whole-slide images from 414 ARISTOTLE participants recruited between October 2011 and July 2018. Tumor cell density (TCD) was defined as the proportion of tumor cells within tumor and stroma tissue, and patients were divided at a cutoff of 0.5 into TCD-high (n=188, 45%) and TCD-low (n=226, 55%) groups. The tool was validated on 116 whole-slide images, achieving a tumor-region precision of 0.873, recall of 0.984 and concordance with manual TCD estimates (Lin's coefficient) of 0.831. Treatment-by-TCD interactions were tested for disease-free and overall survival, with a sensitivity analysis adjusted for KRAS and TP53 status in patients with mutation data.
Treatment effect differed significantly by TCD. The interaction hazard ratio was 0.44 (95% CI 0.24-0.82; P=0.009) for disease-free survival and 0.31 (95% CI 0.16-0.64; P=0.001) for overall survival, and the likelihood-ratio tests favored models including the interaction. In TCD-high tumors, irinotecan-intensified chemoradiotherapy reduced the risk of disease recurrence or death (HR 0.57; 95% CI 0.36-0.90; P=0.014) and of death (HR 0.50; 95% CI 0.30-0.84; P=0.008). In TCD-low tumors there was no benefit (DFS HR 1.29; 95% CI 0.85-1.95; P=0.22) and a nonsignificant trend toward worse overall survival (HR 1.55; 95% CI 0.96-2.51; P=0.07). Across the analyzed cohort, irinotecan had no overall effect (DFS HR 0.89; OS HR 0.92), mirroring the parent trial.
Pathologic complete response in TCD-high tumors was 23% with irinotecan versus 11% with standard chemoradiotherapy (OR 2.46; 95% CI 1.01-5.98; P=0.042), but this was no longer significant after multiple-testing adjustment (P=0.13); in TCD-low tumors it was 14% versus 22%. TCD itself was not prognostic overall, but among patients receiving standard chemoradiotherapy, TCD-high tumors fared worse for DFS (HR 1.54; 95% CI 1.02-2.33) and OS (HR 2.04; 95% CI 1.26-3.29). The interactions held after adjustment for KRAS and TP53 (DFS HR 0.39; OS HR 0.30), although the size of that subset was not reported.
This is a post hoc analysis of a negative trial, and subgroup findings of this kind are prone to chance and overinterpretation. The 0.5 cutoff, the definition of TCD and the analytic plan were not prespecified in ARISTOTLE, and the pathologic response difference did not survive correction for multiple testing. The cohort is restricted to patients with available biopsies, and there is no external validation in an independent trial. The authors themselves describe the result as warranting independent validation before future use in treatment selection. Strengths include randomized treatment allocation, mature survival follow-up and an inexpensive biomarker derived from routine slides.
If confirmed, TCD could provide a practical way to select patients for intensified neoadjuvant therapy using material already collected at diagnosis, and to spare stroma-rich tumors from toxic treatment that may be harmful. Senior author Maria Hawkins said AI-assisted pathology could pinpoint which patients will likely benefit before treatment begins. The immediate next step is validation in other randomized neoadjuvant datasets, particularly those using irinotecan-containing regimens, followed by prospective biomarker-stratified trials.
Source: Shen et al., eBioMedicine 2026;130:106397, published online July 27, 2026. Read the paper
3. Longitudinal adherence to annual stool-based colorectal cancer screening after a negative FIT in four US health systems (PROSPR, Clinical Gastroenterology and Hepatology)
Ethan Halm (Rutgers Robert Wood Johnson Medical School, Rutgers Health), et al., for the NCI PROSPR consortium
Key points
- Observational cohort of 492,812 adults with a negative fecal immunochemical test (FIT), followed for up to 10 years at Parkland Health (Dallas) and Kaiser Permanente Northern California, Southern California and Washington.
- Only 54% were consistent retesters, repeating screening in more than 75% of subsequent years; about 30% repeated at least once and 16% never repeated the test.
- Among patients diagnosed with colorectal cancer, advanced disease at diagnosis occurred in 19% of never-retesters vs 12% in the other two groups.
- Local-stage disease at diagnosis: 56% among consistent retesters vs 44% among never-retesters.
- After adjustment for patient differences, the odds of consistent testing varied more than 20-fold across the four health systems.
Key limitations
- The cohort began with people who had already completed one negative FIT, so it does not describe never-screened people or adherence across all screening modalities.
- Observational design; stage differences may reflect factors other than screening adherence.
- The study period largely preceded widespread use of multitarget stool DNA testing and newly approved blood-based screening.
Future directions
- Identify and spread the organizational practices of systems with high repeat-testing rates (mailed outreach, reminders, navigation).
- Model the outcome cost of inconsistent FIT adherence compared with one-time colonoscopy and newer test intervals.
- Extend longitudinal adherence analyses to stool DNA and blood-based screening tests.
Annual fecal immunochemical testing (FIT) is a widely used, inexpensive and noninvasive colorectal cancer screening strategy, and organized programs in integrated health systems rely heavily on it. Unlike colonoscopy, which can protect for up to a decade, FIT is effective only if repeated regularly: its single-test sensitivity is modest, and its population benefit depends on programmatic, year-after-year participation. Most adherence research has examined uptake of a first test, so far less is known about whether people keep testing after a negative result.
The National Cancer Institute's Population-based Research to Optimize the Screening Process (PROSPR) consortium was created to study the screening process across diverse health systems with linked longitudinal data. Earlier work has shown that outreach strategies such as mailed FIT kits can raise one-time participation, but whether systems sustain repeat testing over many years, and whether lapses matter clinically, has been unclear. This analysis, led by Ethan Halm of Rutgers Robert Wood Johnson Medical School, was designed to describe long-term patterns of repeat FIT use and their association with stage at diagnosis.
The investigators assembled an observational cohort of 492,812 adults who had completed a negative home FIT in one of four health systems: Parkland Health in Dallas, a safety-net system, and Kaiser Permanente Northern California, Southern California and Washington. Patients were followed for up to 10 years after the negative test. Adherence trajectories were classified as consistent retesting (testing in more than 75% of subsequent years), intermittent retesting (at least one repeat) or never retesting. Among patients subsequently diagnosed with colorectal cancer, stage at diagnosis was compared across adherence groups, and variation between health systems was assessed after adjustment for patient characteristics.
Long-term adherence was far from universal. Only 54% of patients repeated the test in more than 75% of subsequent years, about 30% repeated it at least once but irregularly, and 16% never repeated the test after their initial negative result. In practical terms, close to half of a screened population was not following the annual schedule on which the effectiveness of FIT-based programs depends.
Adherence was associated with stage at diagnosis. Among patients who developed colorectal cancer, 19% of never-retesters were diagnosed with advanced disease, compared with 12% in the consistent and intermittent groups, and local-stage disease was found in 56% of consistent retesters versus 44% of never-retesters. Health systems differed strikingly: after adjustment for patient differences, the odds of consistent repeat testing varied more than 20-fold across the four systems, suggesting that organizational factors, rather than patient characteristics alone, drive sustained participation. The study was funded through PROSPR, and no conflicts of interest were disclosed.
The cohort was defined by a negative first FIT, so the findings do not apply to people who have never been screened or to those using other modalities, and adherence to colonoscopy after a positive test was not the focus. The study period largely preceded broad use of combined stool DNA and FIT tests such as Cologuard, and it cannot address blood-based tests. Because the design is observational, the more favorable stage distribution among consistent testers may partly reflect healthier behaviors or better access to care rather than screening alone. Detailed effect estimates for the stage comparisons were not available in the materials we reviewed.
The results argue that FIT-based screening programs should be judged on sustained, not one-time, participation and that health systems have considerable power to improve it. "Systems can have a powerful impact," Halm said, adding that "the best test is the test that gets done." Identifying the practices of high-performing systems, such as automated mailed outreach, reminders and patient navigation, and spreading them to lower-performing settings could translate into earlier-stage diagnoses. Similar longitudinal analyses will be needed for stool DNA and the blood-based tests now entering practice.
Source: Halm et al., "Longitudinal Adherence With Stool-Based Testing for Colorectal Cancer Screening in Four United States Health Systems," Clinical Gastroenterology and Hepatology, July 19, 2026. Read the paper
4. Dietary arginine, codon-dependent MHC-I translation and immunity in colon tumorigenesis (Cell)
Qiushuang Wu (The Rockefeller University), et al., with Heinz-Heinrich Hoffmann (laboratory of Charles Rice); senior author Sohail Tavazoie (Elizabeth and Vincent Meyer Laboratory of Systems Cancer Biology, The Rockefeller University)
Key points
- Preclinical study linking availability of the amino acid arginine to translation of antigen-presentation machinery and to colon tumor formation in mice.
- In cell culture, arginine deficiency left 414 proteins at abnormally low levels, including the products of a trio of HLA genes encoding MHC class I proteins.
- The mechanism proposed is codon-dependent: arginine scarcity impairs translation of transcripts reliant on particular arginine codons, reducing MHC-I.
- Mice fed low-arginine diets developed more colon tumors, whereas those on higher-arginine diets developed fewer.
- The same pathway influenced antiviral immunity: arginine supplementation after infection was tested in mouse influenza and SARS-CoV-2 models.
Key limitations
- Mouse and cell-culture work only; no human dietary or clinical data were reported.
- Sample sizes and effect sizes were not available in the materials we reviewed.
- Arginine has complex, context-dependent effects on tumor and immune metabolism, so dietary manipulation could have unintended consequences.
Future directions
- Test whether arginine status or supplementation affects MHC-I expression and immunotherapy response in human colorectal tumors.
- Examine associations between dietary arginine intake and colorectal neoplasia in epidemiologic cohorts.
- Explore codon-usage-dependent translation as a general mechanism of tumor immune evasion under nutrient stress.
Colorectal cancers develop in an environment strongly shaped by diet, and nutrient availability influences both tumor cells and the immune cells that surveil them. One way tumors escape immune detection is by lowering the display of antigens on MHC class I molecules, which cytotoxic T cells need to recognize malignant cells. How fluctuations in specific dietary nutrients might regulate this antigen-presentation machinery, and thereby tumor formation, has been poorly understood.
Arginine is a conditionally essential amino acid with established roles in T-cell function and in the metabolism of immune and tumor cells. The laboratory of Sohail Tavazoie at The Rockefeller University studies how changes in transfer RNA and codon usage reprogram protein synthesis in cancer. The team asked whether limited availability of a single amino acid could selectively alter translation of specific genes depending on their codon content, and whether this would affect immunity against tumors and viruses.
First author Qiushuang Wu and colleagues used cell cultures to track which proteins changed when arginine was scarce, then turned to mouse models in which animals were fed arginine-restricted or arginine-rich diets and monitored for colon tumor development. In collaboration with Heinz-Heinrich Hoffmann of Charles Rice's laboratory, they extended the work to mouse models of influenza and SARS-CoV-2 infection, including arginine supplementation after infection.
Under arginine deficiency, 414 proteins were present at abnormally low levels. Prominent among them were the products of a trio of HLA genes that encode MHC class I proteins, the molecules responsible for presenting intracellular antigens to cytotoxic T cells. The authors attribute this to codon-dependent translation, in which messenger RNAs that rely on particular arginine codons are translated less efficiently when arginine is scarce, providing a direct link between a dietary nutrient and the immune visibility of cells.
The effect was mirrored in vivo. Mice on low-arginine diets developed more colon tumors, whereas those on higher-arginine diets developed fewer, consistent with improved immune surveillance when arginine was plentiful. The same mechanism appeared relevant to viral infection, which also depends on MHC-I-mediated recognition of infected cells, and the authors examined arginine supplementation after infection in influenza and SARS-CoV-2 models. Quantitative effect sizes and the numbers of animals per experiment were not reported in the materials we reviewed.
The study is entirely preclinical, and the translation of mouse dietary findings to people is uncertain. Arginine also feeds other pathways, including nitric oxide and polyamine synthesis, which can promote tumor growth in some settings, and tumor-associated myeloid cells deplete arginine to suppress T cells; the net effect of supplementation in humans is therefore not predictable from these data. The work does not show that arginine intake alters colorectal cancer risk or immunotherapy response in patients.
The findings nonetheless identify a mechanistic route by which diet could shape antitumor immunity and suggest that nutrient status may modulate the immune visibility of colorectal tumors. Natural next steps include examining MHC-I expression in relation to arginine availability in human colorectal cancers, testing arginine supplementation alongside checkpoint inhibitors in preclinical models, and analyzing dietary arginine intake in epidemiologic cohorts. More broadly, the work adds codon-dependent translation to the list of mechanisms that tumors and pathogens may exploit under nutrient stress.
Source: Wu et al., "Dietary arginine drives codon-dependent MHC-I translation and improves immunity in colon tumorigenesis and respiratory viral infection," Cell, July 30, 2026. Read the paper
5. A DR5 and DNA ligase 3 feedback loop sustaining immunogenicity in mismatch repair-deficient colorectal cancer (Gastroenterology)
Lin Zhang (Keck School of Medicine of USC), S. Hao, Y. Sato, F. Battaglin, K. Ashouri, Z. Liu, X. Lu, J. Yu, Y.-J. Wang, D. Lizardo, R. E. Schoen (Hillman Cancer Center, University of Pittsburgh School of Medicine), J. Xiu and M. Oberley (Caris Life Sciences, Phoenix), with Heinz-Josef Lenz (USC Norris Comprehensive Cancer Center)
Key points
- Laboratory study reported on July 16, 2026, describing a feedback loop between the death receptor DR5 and DNA ligase 3 (Lig3) that maintains immunogenicity in mismatch repair-deficient (dMMR) colorectal cancer.
- Mouse experiments used implanted tumor cells in which a DNA repair gene had been disabled, reproducing a key feature of dMMR tumors.
- The loop is linked to release of extrachromosomal circular DNA (eccDNA), proposed as a signal that alerts immune cells.
- In patient tumor gene-expression data obtained through a collaboration with Caris Life Sciences, tumors with higher DR5 and Lig3 activity responded better to immune checkpoint inhibitors.
- No quantitative effect sizes, cohort sizes or statistical tests were reported in the materials available to us; the authors declared no conflicts of interest and the work was supported by several National Institutes of Health grants.
Key limitations
- Chiefly preclinical; the human component is correlative gene-expression data whose size and analysis were not described in the materials available to us.
- The mouse model depends on engineered loss of DNA repair in implanted cells, which may not reproduce how human dMMR tumors arise and evolve.
- Whether agents that act on DR5 or increase eccDNA release improve immunotherapy outcomes has not been tested.
Future directions
- Test DR5 and Lig3 activity as predictive biomarkers of checkpoint inhibitor benefit in prospective dMMR cohorts.
- Evaluate DR5-directed agents together with checkpoint blockade in preclinical models of colorectal cancer.
- Ask whether the loop can be switched on in mismatch repair-proficient tumors, which rarely respond to immunotherapy.
Mismatch repair-deficient (dMMR) colorectal cancers carry an unusually high mutational load and, as a group, respond well to immune checkpoint inhibitors, which now have a central role in both metastatic and localized disease. Responses are nonetheless not universal, and some dMMR tumors progress from the outset. Clarifying the cellular processes that keep these tumors visible to the immune system may explain that variability and suggest ways to widen the benefit of immunotherapy.
The high neoantigen load of dMMR tumors is the conventional explanation for their immunogenicity, but mutational burden alone does not predict which patients respond. Innate immune sensing of damaged and displaced DNA has emerged as a complementary route to immune activation. Lin Zhang and colleagues at the Keck School of Medicine of USC, with collaborators at the University of Pittsburgh's Hillman Cancer Center and Caris Life Sciences, examined how signaling through the death receptor DR5 interacts with the repair enzyme DNA ligase 3 in tumors lacking mismatch repair.
The team generated mouse models by disabling a DNA repair gene in implanted tumor cells, allowing the pathway to be manipulated in living animals, and paired these experiments with analysis of gene-expression data from patient tumors made available through the Caris Life Sciences collaboration. That second dataset was used to ask whether activity of the pathway tracked with benefit from immune checkpoint inhibitors. Funding came from several National Institutes of Health grants and institutional start-up funds, and the authors reported no conflicts of interest.
The investigators describe a self-reinforcing loop in which DR5 and Lig3 together sustain the immunogenicity of tumor cells lacking mismatch repair. The loop is tied to the release of extrachromosomal circular DNA, small DNA fragments outside the chromosomes that can act as danger signals to innate immune sensors. "We've gathered enough evidence to suggest that this feedback loop is an important piece of the puzzle," Zhang said of the mechanism, positioning it as one reason dMMR tumors draw an immune attack rather than escaping notice.
In the patient dataset, tumors with higher DR5 and Lig3 activity responded better to immune checkpoint inhibitors, which the authors read as evidence that the mechanism may matter clinically. "That suggests we're uncovering a mechanism that could be clinically relevant," Zhang said. The work raises the further possibility that drugs activating the DR5 pathway, or interventions that increase eccDNA release, could strengthen responses to immunotherapy, a question the group says remains under investigation. Response rates by DR5 and Lig3 status, the number of patients analyzed and the statistical methods used were not reported in the materials available to us.
The central evidence comes from mice bearing implanted cells in which DNA repair was switched off, a system that does not reproduce how human dMMR cancers arise, accumulate mutations or shape their microenvironment over time. The human data are observational associations between gene expression and treatment response, which could be confounded by other tumor characteristics, including mutational burden itself. The authors state that more work is needed, including studies in human colorectal tumor cells, before these ideas can be tested in patients, and no therapeutic manipulation of the pathway has yet been attempted in a clinical setting.
Should the findings hold, DR5 and Lig3 activity might help distinguish dMMR tumors likely to respond to checkpoint blockade from those that will not, a distinction that current biomarkers handle poorly. A more ambitious goal would be to induce the loop pharmacologically in tumors that are immunologically quiet, including the mismatch repair-proficient cancers that make up the large majority of colorectal cancer and seldom respond to immunotherapy. Prospective biomarker analyses in treated patients, and preclinical combinations of DR5-directed agents with checkpoint inhibitors, are the natural next steps.
Source: Zhang et al., "A DR5/Ligase 3-mediated feedback loop perpetuates immunogenicity in mismatch repair deficient colorectal cancer," Gastroenterology, reported July 16, 2026. Institutional summary
6. Allogeneic iPSC-derived gamma-delta T cells against patient-derived colorectal tumors in mice (Stem Cell Reports)
Ryoko Futai (Graduate School of Medicine, Kobe University), et al.; senior author Takashi Aoi (Graduate School of Medicine, Kobe University)
Key points
- Preclinical study published July 23, 2026, in which gamma-delta T cells were reprogrammed into induced pluripotent stem cells (iPSCs) and then differentiated back into gamma-delta T cells suitable for allogeneic use.
- The manufacturing process used no animal cells or animal-derived extracts, a requirement for clinical application.
- An overall expansion of about 80,000-fold was achieved reproducibly, the key step toward mass production.
- The cells were tested against colorectal tumors derived from two patients and implanted in mice, given either close to the tumor or intravenously; each experiment used three or four mice.
- With intravenous dosing one week after tumor implantation, tumor weights fell by 43% to 92% in the three mice tested.
Key limitations
- Very small experiments, with three or four mice per condition and tumors from only two patients.
- No comparator cell product or standard treatment, and no safety or persistence data in the materials available to us.
- The investigators state the approach is preclinical and not at a stage where it can be used in patients.
Future directions
- Test the cells against a wider panel of patient-derived colorectal tumors, including microsatellite-stable disease.
- Combine the platform with chimeric antigen receptor engineering to sharpen tumor targeting.
- Move to clinical-grade manufacture and first-in-human safety evaluation.
Cell therapies have transformed the treatment of several blood cancers but have made little headway in solid tumors, colorectal cancer included. Most approved products are autologous, manufactured individually from each patient's own lymphocytes, which is slow, expensive and dependent on the quality of T cells in people who have often had extensive chemotherapy. An off-the-shelf product that could be made in bulk, stored and given to many recipients would remove those constraints, provided the cells do not attack the recipient's tissues.
Gamma-delta T cells are an unconventional T-cell population that recognizes stressed and transformed cells largely without the classical restrictions that cause graft-versus-host disease, which makes them attractive for use across donor and recipient pairs. Their development as therapy has been limited by the difficulty of obtaining them in sufficient numbers. Induced pluripotent stem cells offer a potentially unlimited and standardized starting material, and the group of Takashi Aoi at Kobe University set out to produce clinically compatible gamma-delta T cells from iPSCs and test them against human tumors.
First author Ryoko Futai and colleagues reprogrammed gamma-delta T cells into iPSCs and then differentiated them back into gamma-delta T cells, deliberately avoiding animal cells and animal-derived extracts at every step because such materials are not acceptable for clinical products. The resulting cells were tested against colorectal tumors taken from two patients and implanted into mice. Two routes of administration were compared: delivery close to the tumor, and intravenous delivery one week after the tumor was implanted. Each experiment used three or four animals.
The process achieved an overall expansion of roughly 80,000-fold and did so reproducibly, which the authors present as evidence that the approach could be scaled for mass production rather than remaining a laboratory curiosity. The resulting cells attacked and shrank the patient-derived colorectal tumors growing in mice, which the group describes as the first demonstration of this effect for iPSC-derived gamma-delta T cells against patient tissue.
Activity was not limited to local delivery. When the cells were infused intravenously one week after tumor implantation, tumor weights fell by between 43% and 92% in the three animals tested, a result Futai described as suggesting potential for future systemic therapy. Local administration near the tumor was also effective. The published report contains the detailed protocols, dosing and statistical analysis; the summary materials available to us give no information on control groups, persistence of the transferred cells or toxicity.
The experiments are very small. Three or four mice per condition and tumors from only two patients cannot capture the heterogeneity of colorectal cancer, where mismatch repair status, sidedness, driver mutations and stromal content all influence how tumors respond to immune attack. Without a comparator product it is impossible to say how these cells perform against existing approaches, and the absence of reported safety, persistence and trafficking data leaves central questions about allogeneic cell therapy unanswered. The investigators are explicit that the work is preclinical.
The contribution here is less a therapeutic result than a manufacturing one: an animal-component-free process that expands a difficult cell type roughly 80,000-fold and yields cells with demonstrable activity against human tumor tissue in vivo. Scaling to a broader panel of patient-derived tumors, adding chimeric antigen receptors to improve targeting of colorectal antigens, and establishing good manufacturing practice production would be the logical sequence toward a first-in-human study. Aoi's group has said it hopes that combining this platform with CAR engineering could open new options in solid tumors.
Source: Futai et al., "Allogeneic iPSC-derived gamma-delta T cells demonstrate antitumor efficacy against patient-derived tissues," Stem Cell Reports, July 23, 2026. Read the paper · Institutional summary
7. Sexual orientation and gender identity disparities in colorectal, cervical and breast cancer screening in the United States (Cancer)
L. Arena (The Ohio State University Wexner Medical Center, Columbus), Q. Alizai, A. Elemosho, O. P. Chatzipanagiotou, and senior author Timothy M. Pawlik (The Ohio State University Wexner Medical Center and James Comprehensive Cancer Center)
Key points
- Retrospective analysis of 2018-2022 Behavioral Risk Factor Surveillance System (BRFSS) data covering 663,924 unweighted respondents eligible for colorectal, cervical or breast cancer screening under USPSTF criteria; 1.2% were sexual orientation minorities and 0.4% gender identity minorities.
- Survey-weighted Poisson regression adjusted for age, race, income, education, marital status, employment, insurance and rurality.
- Colorectal screening: no overall difference for sexual orientation minority women; adjusted prevalence ratio (aPR) 1.10 (95% CI 1.00-1.21) for sexual orientation minority men versus heterosexual men; no overall gender identity disparity.
- In exploratory analysis of female-to-male transgender respondents, colorectal screening aPR was 0.20 (95% CI 0.04-0.92) and cervical screening aPR 0.50 (95% CI 0.30-0.84).
- Other screening gaps: cervical aPR 0.92 (95% CI 0.87-0.97) and breast aPR 0.84 (95% CI 0.77-0.92) for sexual orientation minority women; cervical aPR 0.58 (95% CI 0.37-0.90) and breast aPR 0.24 (95% CI 0.08-0.74) for gender identity minority women.
Key limitations
- Cross-sectional, self-reported survey data, subject to recall and social desirability bias.
- The sexual orientation and gender identity module was optional and state-administered, reducing sample size and statistical power for subgroups; the transgender exploratory estimates have very wide confidence intervals.
- A single transgender-identity question did not capture sex assigned at birth, so anatomy-appropriate screening eligibility may be misclassified.
Future directions
- Adopt two-step sexual orientation and gender identity measures in national surveys and cancer registries.
- Base insurance coverage of screening on anatomy rather than administrative gender markers.
- Test affirming screening interventions, including self-collected sampling, in transgender and gender-diverse populations.
Colorectal cancer screening works only to the extent that eligible people actually undergo it, and participation is distributed unevenly across the population. Groups that encounter discrimination or discomfort in clinical settings, or whose records do not match the organ-based eligibility rules embedded in screening programs, may be systematically missed. Sexual and gender minority adults have long been identified as such a group, but national estimates have been limited by surveys that did not ask about sexual orientation or gender identity.
The Behavioral Risk Factor Surveillance System, a large annual telephone survey run by US states, has added an optional module on sexual orientation and gender identity, making population-level comparisons possible for the first time at scale. Prior work has suggested lower uptake of cervical and breast screening among sexual and gender minority women, with much less information about colorectal cancer, where screening eligibility does not depend on sex and might therefore be expected to show smaller gaps. Investigators at The Ohio State University Wexner Medical Center set out to quantify disparities across all three screening programs in one national dataset.
The team analyzed BRFSS responses from 2018 to 2022, restricted to adults eligible for colorectal, cervical or breast cancer screening under US Preventive Services Task Force criteria, yielding 663,924 unweighted respondents, of whom 1.2% reported a minority sexual orientation and 0.4% a minority gender identity. Screening receipt was compared using survey-weighted Poisson regression adjusted for age, race, income, education, marital status, employment, insurance and rurality, producing adjusted prevalence ratios. A Blinder-Oaxaca decomposition was used to apportion the cervical screening gap between measured and unmeasured factors, and an exploratory analysis examined female-to-male transgender respondents separately.
For colorectal cancer screening, the overall picture was reassuring: sexual orientation minority women did not differ from heterosexual women, sexual orientation minority men were if anything slightly more likely to be screened (aPR 1.10; 95% CI 1.00-1.21), and no overall gender identity disparity was detected. The exploratory subgroup analysis told a different story. Among female-to-male transgender respondents, colorectal screening was markedly lower (aPR 0.20; 95% CI 0.04-0.92), as was cervical screening (aPR 0.50; 95% CI 0.30-0.84), although the very wide confidence interval for the colorectal estimate reflects a small number of respondents.
Disparities were clearer in the sex-specific programs. Sexual orientation minority women were less likely than heterosexual women to receive cervical (aPR 0.92; 95% CI 0.87-0.97) and breast (aPR 0.84; 95% CI 0.77-0.92) screening, and the gaps were far larger for gender identity minority women (cervical aPR 0.58; 95% CI 0.37-0.90; breast aPR 0.24; 95% CI 0.08-0.74). The decomposition analysis indicated that measured factors, chiefly insurance, age and marital status, explained about 64% of the cervical screening gap for sexual orientation minority women, leaving more than a third unexplained by the variables available. Self-reported cancer prevalence did not differ by sexual orientation or gender identity status for any of the three cancers.
The analysis inherits the weaknesses of its data source. BRFSS is cross-sectional and entirely self-reported, so screening receipt may be misremembered or misreported, and the sexual orientation and gender identity module was optional and administered only by some states, limiting both sample size and generalizability. Because sexual orientation and gender identity were captured in separate items, analytic denominators differ between analyses, and the single transgender-identity question did not record sex assigned at birth, so eligibility for anatomy-specific screening may be misclassified. The transgender subgroup estimates, the most striking in the paper, rest on small numbers.
The practical message for colorectal cancer screening is that headline comparisons can obscure the groups most at risk of being missed: aggregate rates among sexual and gender minority adults were close to those of the comparison groups, yet transgender respondents in the exploratory analysis were far less likely to be screened. The authors recommend two-step measures of sexual orientation and gender identity in national surveys and cancer registries, insurance policies that key coverage to anatomy rather than administrative gender markers, and research on affirming screening approaches such as self-collected sampling. These are implementation questions, well suited to pragmatic trials within health systems.
Source: Arena et al., "Sexual orientation and gender identity based disparities in colorectal, cervical, and breast cancer screening in the United States," Cancer 2026;132(14):e70462 (covered July 16, 2026; exact online date not verified). Read the paper