Microbiome-derived TOPOSCORE is associated with response to immune checkpoint inhibitor re-induction after chemotherapy in primary resistant metastatic melanoma

Abstract Background Primary resistance to immune checkpoint inhibitors (ICIs) remains a major challenge in metastatic melanoma. The prospective multicenter PROMIT trial showed that intermittent chemotherapy could restore sensitivity to ICI re-induction in a subset of patients. We investigated whether gut microbiome ecosystem states characterized by TOPOSCORE were associated with tumor response in this setting. Methods In the phase II PROMIT trial (NCT04225390), 24 patients with longitudinal stool samples were included in this exploratory microbiome analysis. ICI-resistant patients received two cycles of dacarbazine followed by ICI re-induction. Stool samples underwent shotgun metagenomic sequencing. Microbiome ecosystem states were classified into SIG1- or SIG2-associated ecosystem states using the TOPOSCORE framework, which is based on species-interacting groups (SIGs) defined by ecological co-abundance patterns. Results Baseline TOPOSCORE distributions and s-scores were not associated with clinical outcome. In the nine patients with complete BL-CTX-ICI sampling, six retained the same TOPOSCORE category throughout treatment and three changed category. Of the two patients with PR, one remained SIG2 throughout, whereas the other changed from SIG2+ to SIG1+ after chemotherapy and returned to SIG2+ during ICI re-induction. Analyses using all available samples at each phase likewise did not show significant differences in s-score among PR, SD, and PD groups. Conclusions Baseline TOPOSCORE classification was insufficient to predict treatment outcome in this cohort. Longitudinal TOPOSCORE patterns were heterogeneous, and the small number of complete longitudinal profiles precludes conclusions that chemotherapy induced a favorable microbiome transition. These exploratory findings support further evaluation of microbiome ecosystem dynamics in larger prospective cohorts.