Microbiome Signatures Predict Melanoma Recurrence After Adjuvant ICI

Emerging data from a large translational analysis suggest that baseline gut microbiome composition may predict recurrence-free survival (RFS) in patients with resected high-risk melanoma who receive adjuvant immune checkpoint inhibitors (ICIs), potentially offering a clinically actionable biomarker to guide treatment selection.¹

In the study, published in Cell, investigators analyzed stool samples from 674 patients enrolled in CheckMate 915 (NCT03068455), comparing microbiome features with recurrence outcomes across geographically diverse populations. The findings demonstrate that specific bacterial taxa, particularly within Eubacterium, Ruminococcus, Firmicutes, and Clostridium, were associated with recurrence risk, and that predictive performance improved substantially when accounting for interindividual microbiome similarity.

“These findings are very reliable, and it’s really convincing,” said Jiyoung Ahn, PhD, professor and Associate Center Director for Population Science at the NYU Perlmutter Cancer Center, a senior investigator on the study. “I hope this research will help patients and clinicians guide therapy decisions.”

Predictive Biomarkers in an Unmet Need Setting

Adjuvant ICI has transformed outcomes in resected stage III and IV melanoma, yet approximately 25% to 40% of patients still experience recurrence, underscoring the need for predictive biomarkers.¹ Current clinical and pathologic features incompletely stratify risk, particularly in the context of expanding immunotherapy use in earlier-stage disease.

“Immunotherapy has been a breakthrough treatment…but this therapy doesn’t really work all the time,” Ahn noted. “Only a subset—20% to 40%—respond. So it’s a really big question: can we identify biomarkers before initiation of therapy?”

The present study sought to address limitations of prior microbiome research, which has been constrained by small sample sizes and inconsistent reproducibility across cohorts. By leveraging a large, randomized trial with standardized biospecimen collection, investigators aimed to define robust microbial predictors of recurrence after treatment with nivolumab (Opdivo) and ipilimumab (Yervoy) in the adjuvant setting.² Ahn said predicting risk of recurrence is crucial as adjuvant ICI is being used more widely.

Geographic Variation and Microbiome Context

A key finding was the strong influence of geographic variation on microbiome composition. Patients were enrolled across North America, Western and Eastern Europe, Australia, and other regions, and microbiome profiles differed significantly by location.¹

“We noticed that microbiome profiles reflect environment and diet,” Ahn explained. “So microbiome profiles are different [across] geographic areas…we thought maybe we need a large study encompassing different geographic areas to develop reliable biomarkers.”

Region-specific analyses identified distinct bacterial taxa associated with recurrence, but cross-regional generalizability was initially limited. Investigators addressed this by incorporating microbiome similarity metrics, specifically Jensen-Shannon divergence, to match patients across regions based on overall microbial composition.

“When people are similar in their backgrounds, with overall microbiome [that are] similar, the predictability was very high,” Ahn said. “If your background was less close, and microbiome diversity is substantially different, predictability tended to decrease.”

This approach markedly improved predictive performance, with area under the curve (AUC) values ranging from 0.78 to 0.94 among closely matched individuals whose divergence was 0.11 or less.¹ These findings suggest that microbiome-informed risk stratification may require contextualization within broader microbial ecosystems rather than reliance on a single taxon.

Region-Specific Risk Analyses

In North American patients, 9 bacterial taxa were associated with recurrence, several of which had previously been associated with recurrence along with Aeromonas salmonicida and Peptostreptococcus anaerobius being identified as novel markers. Similar findings showed particular taxa in other regions as well.

A meta-analysis of the region-specific markers found 7 recurrence-associated taxa that were also linked in cross-region analyses, though several were found to be not significant across regions.

Stability of the Microbiome During Treatment

Importantly, the study demonstrated that gut microbiome composition remained largely stable during ICI treatment, based on longitudinal sampling at baseline, 7 weeks, and 29 weeks. The correlations with recurrence were also shown to be consistent at each time point.

“We also observed that these bacteria remained stable during immunotherapy treatment, which is great,” Ahn said. “We can see before initiation of treatment…these bacteria [are] a reliable predictor.”

This temporal stability enhances the clinical feasibility of microbiome-based biomarkers, as a single pretreatment sample may suffice for risk assessment. It also supports the hypothesis that baseline microbial composition influences immune response dynamics rather than being substantially altered by therapy itself.

Mechanistic and Translational Implications

Functional analyses linked recurrence-associated taxa to metabolic pathways, particularly 8 involved in carbohydrate metabolism in North American patients, although these were not linked to recurrence in the other regions.¹ Two pathways related to starch and sucrose metabolism and glycerolipid metabolism were associated with protection from recurrence at a relaxed threshold, though they failed to meet significance at P < .05. These findings align with prior evidence suggesting that diet and microbial metabolites may modulate antitumor immunity.

“We found that several fiber-metabolizing bacteria… strongly predicted melanoma recurrence,” Ahn said, highlighting a potential link between diet, microbiome composition, and treatment outcomes.

The study adds to a growing body of literature demonstrating that the gut microbiome influences ICI efficacy, including preclinical and early clinical data supporting fecal microbiota transplantation and dietary interventions as modulators of response.

Investigators also looked at grade 3 or higher immune-related adverse events, finding no link to gut microbiome composition in all patients or in individual regions. They reported a link between certain taxa and reduced occurrence of these toxicities in the Eastern European cohort, but stated that the correlation was weak due to the low rate of high-grade events.

Looking ahead, Ahn emphasized ongoing efforts to expand beyond bacterial profiling. “Gut microbiome has other components—fungi, viruses,” she said. “We are currently working… to see whether fungi and virus are important predictors.”

Clinical Applications and Future Directions

Although the findings are not yet ready for routine clinical use, they suggest several potential applications. Baseline microbiome profiling could enable risk stratification prior to adjuvant therapy, informing discussions about expected benefit and alternative strategies.

“If we can efficiently tell the responsiveness before initiation of therapy, that will be really great,” Ahn said. “It saves time and effort, and we can make informed decisions and lead to personalized therapy.”

However, she cautioned against overinterpretation in the near term. “This is the first study—[it] needs to be replicable… and applied to other melanoma trials,” she noted.

However, it has the potential to be applied more broadly to other solid tumors like lung cancer or head and neck cancer, she said. “In many other cancers, we now use immunotherapy, so this modeling method can be widely applicable for other cancers.”

Further studies are needed to validate these biomarkers in independent datasets and to integrate microbiome data with clinical, genomic, and immunologic variables.

Conclusion

This large-scale analysis of the CheckMate 915 cohort provides compelling evidence that baseline gut microbiome composition is associated with melanoma recurrence following adjuvant ICB. By accounting for geographic and compositional variability, investigators demonstrated that microbiome-based models can achieve high predictive accuracy, supporting their potential as clinically actionable biomarkers. They could also help demonstrate that dietary changes, probiotics, and fecal microbiota transplants are effective tools for enhancing response to ICI.^3,4

As immunotherapy continues to expand in earlier-stage melanoma, such tools may play an increasingly important role in personalizing treatment strategies and improving patient outcomes.

REFERENCES

1. Usyk M, Hayes RB, Knight R, et al. Gut microbiome is associated with recurrence-free survival in patients with resected high-risk melanoma receiving adjuvant immune checkpoint blockade. Cell. 2026;189:1-10. doi:10.1016/j.cell.2026.03.041

2. Weber JS, Schadendorf D, Del Vecchio M, et al. Adjuvant therapy of nivolumab combined with ipilimumab versus nivolumab alone in patients with resected stage IIIB-D or stage IV melanoma (CheckMate 915). J Clin Oncol. 2023;41:517-527. doi:10.1200/JCO.22.00533

3. Spencer CN, McQuade JL, Gopalakrishnan V, et al. Dietary fiber and probiotics influence melanoma immunotherapy response. Science. 2021;374:1632-1640. doi:10.1126/science.aaz7015

4. Baruch EN, Youngster I, Ben-Betzalel G, et al. Fecal microbiota transplant promotes response in immunotherapy-refractory melanoma. Science. 2021;371:602-609. doi:10.1126/science.abb5920