Gut microbiome functional pathways outperform taxonomic profiles in predicting immune checkpoint inhibitor response in non-small cell lung cancer: an interpretable machine learning approach with SHAP - Summary - MDSpire

Gut microbiome functional pathways outperform taxonomic profiles in predicting immune checkpoint inhibitor response in non-small cell lung cancer: an interpretable machine learning approach with SHAP

  • By

  • Feifei Wei

  • Yoshiro Nakahara

  • Junya Isobe

  • Yuka Igarashi

  • Haruhiro Saito

  • Shuji Murakami

  • Tetsuro Kondo

  • Hidetomo Himuro

  • Taku Kouro

  • Tomoya Matsui

  • Satoshi Wada

  • Takuya Tsunoda

  • Kiyoshi Yoshimura

  • Tetsuro Sasada

  • May 15, 2026

Share

Objective:

To evaluate the predictive significance of taxonomic versus functional gut microbiome characteristics on immune checkpoint inhibitor (ICI) therapy outcomes in non-small cell lung cancer (NSCLC).

Key Findings:
  • Functional profiles from MetaCyc pathways showed the strongest correlation with treatment response, indicating their predictive power.
  • A signature of four specific pathways was identified as highly predictive of treatment outcomes.
  • Key factors influencing responder classification were linked to nitrogen metabolism and short-chain fatty acid biosynthesis, highlighting their importance.
Interpretation:

Functional microbiome profiles provide a more accurate reflection of interactions between the microbiome and host, suggesting potential for precision interventions focused on metabolic pathways, which may enhance treatment efficacy.

Limitations:
  • Study limited to a specific ethnic group (Japanese individuals), which may affect generalizability to other populations.
  • Sample size of 77 may limit the robustness of findings and their applicability.
Conclusion:

Functional characteristics of the gut microbiome are superior to taxonomic features in predicting responses to ICIs in NSCLC, highlighting the importance of metabolic pathways in treatment outcomes.

Original Source(s)

Related Content