Immune checkpoint blockade (ICB) has significantly improved outcomes for patients with advanced-stage cancers, including melanoma. However, ICB's efficacy varies greatly among patients, and treatment can cause immune-related adverse events (irAEs), such as colitis. While cross-sectional studies have explored the relationship between the gut microbiome and ICB response, they lack consensus regarding microbial biomarkers, likely due to methodological, biological, and clinical confounders, and the inherent variability of the gut microbiome. Numerous microbiome-targeted clinical trials, including fecal microbiota transplantation (FMT) trials, are underway to enhance ICB efficacy; however, understanding the longitudinal dynamics of the gut microbiome during ICB treatment is crucial to interpreting these trials and improving treatment strategies. This study addresses this knowledge gap by conducting a longitudinal analysis of gut microbiome changes in melanoma patients receiving ICB.

The existing literature shows a correlation between gut microbiome composition and ICB response in melanoma patients, but results have been inconsistent across studies. Several studies have identified microbial taxa associated with either improved or diminished response to ICB, often focusing on baseline microbiome profiles. However, a comprehensive understanding of how the gut microbiome changes over time during ICB treatment remains limited. Prior studies highlight the importance of short-chain fatty acid (SCFA)-producing bacteria and their association with positive ICB outcomes. Conversely, bacteria linked to chronic inflammatory diseases have been associated with poorer responses. The lack of longitudinal studies has hampered a more complete picture of the dynamic interplay between the gut microbiome and ICB efficacy.

This multicenter study enrolled 175 patients with unresectable stage 3 or 4 cutaneous melanoma treated with ICB (anti-PD-1 alone or in combination with anti-CTLA-4). Shotgun metagenomics was employed to profile the gut microbiome at four time points (baseline and three subsequent visits over 12 weeks). MetaPhlAn4 and HuMAnN3 were used for taxonomic and metabolic pathway analyses. Bayesian regression models with higher-order interactions were used to analyze the longitudinal changes in SGBs and pathways in patients with PFS ≥12 months versus <12 months, adjusting for confounders such as treatment regimen, irAEs (including colitis), PPI use, antibiotic use, prior BRAF/MEK inhibition, and cancer center. A log ratio (balance) was constructed from SGBs consistently associated with each PFS group to evaluate its predictive value for PFS≥12 and OS. Post-hoc contrasts were performed to analyze microbial dynamics in different clinical contexts. The study also validated its findings using data from six independent melanoma cohorts.

The study revealed distinct longitudinal microbiome patterns between patients with PFS ≥12 months and those with PFS <12 months. While several SGBs differed only at baseline, others showed differential abundance only after ICB initiation. Five SGBs consistently showed higher abundance in patients with PFS ≥12 months (*Agathobaculum butyriciproducens*, *Intestinibacter bartlettii*, *Dorea* sp., *Lactobacillus gasseri*, *Lacrimispora celerecrescens*), many of which are SCFA producers. Conversely, four SGBs (*Ruthenibacterium lactatiformans*, *Prevotella copri*, *Ruminococcaceae* unclassified, and an unidentified Bacteroidetes SGB) were consistently higher in patients with PFS <12 months. A log ratio constructed from these SGBs predicted OS (HR=1.67, P=0.035). The study also observed different microbial dynamics in the context of ICB regimens (monotherapy vs. combination therapy), colitis development, and PPI use. Certain SGBs exhibited opposite abundance patterns in patients with PFS ≥12 months depending on the treatment regimen. For example, *Coprococcus eutactus*, *Butyricicoccus* sp., and *Parabacteroides merdae* displayed divergent patterns between monotherapy and combination therapy. The study also identified a baseline microbial balance predictive of colitis development (AUC=0.723, P=0.00055).

This study provides novel insights into the dynamic interplay between the gut microbiome and ICB response in advanced melanoma. The longitudinal analysis demonstrated that changes in the gut microbiome during ICB treatment, rather than solely baseline composition, are critical for understanding treatment response. The identified SGBs and pathways associated with improved or diminished response provide potential targets for microbiome-based interventions. The observation of distinct patterns according to the ICB regimen, presence of colitis, and PPI use underscores the importance of considering these factors when designing and interpreting microbiome-targeted therapies. The findings support the use of longitudinal monitoring of the gut microbiome to assess ICB efficacy and irAEs.

This study highlights the dynamic nature of the gut microbiome's response to ICB in advanced melanoma, emphasizing the need for longitudinal analysis. The identified microbial signatures and predictive balances offer valuable insights for developing and implementing personalized microbiome-based strategies. Future research should focus on larger-scale studies incorporating multi-omics data to further elucidate the mechanisms driving these associations and to optimize microbiome-modulating therapies.

This study has several limitations, including the simplification of microbial dynamics to linear trajectories and potential comparability issues with previous studies due to differences in taxonomic databases. The smaller number of samples in some post-hoc comparisons may limit the generalizability of certain findings. Furthermore, the study did not consider all potential confounders that might influence microbiome composition and ICB response, and the observed associations require further validation in independent cohorts.