1265 Associations of the gut microbiome with immune checkpoint inhibitor response across different cancer types and cohorts identified using random forest and language models
Year: 2024
Type: conference-abstract
Abstract: Background Immune checkpoint inhibitors (ICIs) are effective against various cancer types but only benefit a subset of patients. The gut microbiome is one of the potential tumor-extrinsic predictive biomarkers for ICI efficacy, but cohort differences have led to inconsistencies in microbiome characteristics associated with ICI benefits. Methods We collected 942 metagenomic datasets from ICI-treated cancer patients with melanoma, non-small cell lung cancer and renal cell carcinoma across nine public cohorts. The datasets were analyzed using the Ez-Mx platform and the EzBioCloud database. To predict ICI response, we utilized random forest (RF) and a RoBERTa-based language model (LM). The LM was pretrained on 13,418 quality-controlled taxonomic profiles and fine-tuned for ICI response prediction, using species names sorted by abundance as tokens. Model performance was assessed using three-fold cross-validation (CV) and leave-one-out cross-validation (LOOCV), each repeated 100 times. Results We compared different response criteria and found that excluding stable disease samples improved AUROC values compared to including them as either responders or non-responders in both CV (0.653, 0.583 and 0.637) and LOOCV (median: 0.596, 0.573 and 0.575). Taxonomic profiles at the species level achieved the highest AUROC compared to other taxonomic ranks. Therefore, we utilized species-level profiles without stable disease samples for RF and LM. The LM generally performed better than the RF, with higher AUROC in Derosa_2022_NSCLC1 (0.639 and 0.611), McCulloch_2022_MEL2 (0.618 and 0.564), Routy_2018_NSCLC3 (0.676 and 0.660), Routy_2018_RCC3 (0.658 and 0.560) and Spencer_2021_MEL4 (0.604 and 0.554) in LOOCV. Conversely, the RF outperformed the LM in Frankel_2017_MEL5 (0.551 and 0.635), Matson_2018_MEL6 (0.305 and 0.384) and Peters_2019_MEL7 (0.744 and 0.801). In Lee_2022_MEL,8 both models had similar performances (0.576 and 0.580). We identified key features significantly contributing to the prediction, with species from the Lachnospiraceae family being particularly important. Blautia wexlerae, Anaerostipes hadrus and Fusicatenibacter saccharivorans were the most important, enriched in the responders with p-values of 0.009, 0.053 and 0.016, respectively, according to the random-effects meta-analysis. Conclusions We explored the relationships between the gut microbiome and ICI response across different cancer types and cohorts using RF and LM. Despite the complexity of ICI response prediction, our models achieved reasonable AUROC values, indicating potential utility in clinical applications. We identified bacterial species linked to ICI response, suggesting their roles in immune modulation. Understanding the relationships between the gut microbiome, the immune system, and cancer may be crucial for comprehending the complexity of ICI response. References Derosa L, et al. Intestinal Akkermansia muciniphila predicts clinical response to PD-1 blockade in patients with advanced non-small-cell lung cancer. Nat Med. 2022;28:315–324. McCulloch JA, et al. Intestinal microbiota signatures of clinical response and immune-related adverse events in melanoma patients treated with anti-PD-1. Nat Med. 2022;28:545–556. Routy B, et al. Gut microbiome influences efficacy of PD-1–based immunotherapy against epithelial tumors. Science. 2018;359:91–97. Spencer CN, et al. Dietary fiber and probiotics influence the gut microbiome and melanoma immunotherapy response. Science. 2021;374:1632–1640. Frankel AE, et al. Metagenomic shotgun sequencing and unbiased metabolomic profiling identify specific human gut microbiota and metabolites associated with immune checkpoint therapy efficacy in melanoma patients. Neoplasia. 2017;19:848–855. Matson V, et al. The commensal microbiome is associated with anti–PD-1 efficacy in metastatic melanoma patients. Science. 2018;359:104–108. Peters BA, et al. Relating the gut metagenome and metatranscriptome to immunotherapy responses in melanoma patients. Genome Med. 2019;11:1–14. Lee KA, et al. Cross-cohort gut microbiome associations with immune checkpoint inhibitor response in advanced melanoma. Nat Med. 2022;28:535–544.
Authors: Junwon Yang, Hyun-Seok Oh, Taehee Lee, Ahin Lee, Mauricio Chalita, Sein Park, Yeonjae Jung, Jeongsup Moon, Laura Kang, Jongsik Chun
Institution: Seoul National University
Language: English
Subfield: Oncology
Field: Medicine
Domain: Health Sciences
Open Access status: gold
Estimated APC paid: $2,490
OpenAlex ID: W4404064332