EGFR-TKIs are the standard treatment for EGFR-mutant NSCLC, but biomarkers for predicting response and toxicity are lacking. EGFR mutations occur in approximately 50% of Asian NSCLC patients, most commonly as exon 19 deletions or L858R substitutions. EGFR tyrosine kinase inhibitors are highly effective first-line therapies, yet biomarkers that predict efficacy and identify patients at risk for severe adverse events remain an important unmet clinical need.
The gut microbiota has emerged as a potential biomarker for cancer treatment outcomes. Growing evidence supports a bidirectional lung-gut axis through which intestinal microbiota can influence pulmonary immunity and inflammatory responses via microbial metabolites including short-chain fatty acids and bile acids, and through systemic cytokine signaling. Conversely, lung tumors and their treatment may disrupt intestinal homeostasis.
Gut microbiota composition has been linked to immunotherapy response in NSCLC patients without EGFR mutations, with specific taxa such as Akkermansia muciniphila and Bifidobacterium breve associated with favorable outcomes from immune checkpoint inhibitors. However, no prior study had examined gut microbiota as a biomarker specifically in EGFR-mutant NSCLC patients receiving targeted therapy.
Diarrhea is one of the most common and clinically significant side effects of EGFR-TKIs, occurring with variable severity across patients. Understanding whether pretreatment microbiota composition predicts diarrhea severity could enable prophylactic interventions, while identifying microbiota signatures associated with EGFR-TKI efficacy could improve patient selection and treatment personalization.
A prospective observational study enrolling 21 treatment-naive EGFR-mutant NSCLC patients. The study was conducted at Hirosaki University Hospital between July 2020 and July 2023. All patients had advanced NSCLC with confirmed EGFR mutations scheduled for first-line EGFR-TKI therapy. Patients with inflammatory bowel disease, active infectious diseases, or poorly controlled diabetes were excluded. None had taken antibiotics before sample collection.
Fecal samples were collected prior to EGFR-TKI initiation using a standardized kit with guanidine preservation solution. DNA was extracted using the DNeasy PowerSoil Pro Kit, and 16S rRNA gene amplification targeted the V1-V2 region using primer set 27Fmod/338R. Paired-end sequencing was performed on an Illumina MiSeq platform with 251-bp reads and MiSeq Reagent v2.
Bioinformatics processing used the DADA2 pipeline for merging, filtering, and denoising sequences. Taxonomic assignment was performed using the QIIME2 feature-classifier with the Greengenes 13_8 database. Diversity analyses included the Shannon index for alpha-diversity (species richness and evenness within samples) and UniFrac distances for beta-diversity (compositional differences between samples), assessed by principal coordinate analysis.
Linear discriminant analysis Effect Size (LEfSe) was applied to identify specific taxa most responsible for differences between the partial response group and the stable or progressive disease group. Best overall response was evaluated per RECIST v1.1 criteria. Diarrhea severity was graded using CTCAE v5.0, with grade 2 or higher defined as clinically significant toxicity.
Higher gut microbiota diversity was associated with milder diarrhea. Of the 21 patients, seven developed grade 2 or higher diarrhea during EGFR-TKI treatment, while 14 experienced no or grade 1 diarrhea. The Shannon alpha-diversity index was significantly higher in the low-severity diarrhea group compared to the high-severity group (p = 0.0367), suggesting that a more diverse pretreatment gut microbiome may protect against severe EGFR-TKI-induced intestinal toxicity.
In terms of treatment efficacy, 11 patients (52.4%) achieved a partial response, nine (42.8%) had stable disease, and one (4.8%) had progressive disease. No patient achieved a complete response. Beta-diversity analysis showed no significant difference in unweighted UniFrac distances between the partial response and stable or progressive disease groups (p = 0.118), but weighted UniFrac distances differed significantly (p = 0.041), indicating that the relative abundance of gut microbial taxa differs meaningfully between responders and non-responders.
LEfSe analysis identified Ruminococcus as the genus most discriminating between the partial response group and the stable or progressive disease group. Univariate analysis confirmed a highly significant difference in Ruminococcus relative abundance between the two groups (p = 0.0018). ROC curve analysis identified a Ruminococcus abundance cutoff of 1.4% of total flora, above which EGFR-TKI response was more likely (AUC 0.96, p less than 0.001).
Among non-responders in the stable or progressive disease group, Fusobacteria were detected in three patients at abundances of 0.69-5.32% of total flora. All three patients with detectable Fusobacteria had poor EGFR-TKI responses, consistent with prior evidence linking Fusobacteria to malignancy, unfavorable prognosis, and treatment failure in other cancer types.
Ruminococcus abundance as an efficacy biomarker extends prior immunotherapy findings. Previous studies have shown that Ruminococcus abundance is associated with durable clinical benefit and enhanced anti-tumor effects from immune checkpoint inhibitors in other cancer types. The current study extends this biomarker signal to EGFR-TKI therapy in EGFR-mutant NSCLC, suggesting that Ruminococcus may modulate therapeutic efficacy through the gut-lung axis regardless of whether treatment is immunotherapy or targeted therapy.
Ruminococcus species play important functional roles in the gut including short-chain fatty acid production, bile acid metabolism, and maintenance of mucosal integrity. These functions may influence systemic immune status, drug metabolism, and mucosal protection in ways that ultimately affect EGFR-TKI efficacy. However, the 16S V1-V2 sequencing approach used here cannot resolve species-level distinctions between Ruminococcus gnavus, Ruminococcus bromii, and other members of this genus with potentially different functional profiles.
The overall phylum-level composition observed in this cohort -- dominated by Firmicutes, Actinobacteria, and Bacteroidota, accounting for over 90% of the microbiome -- was consistent with prior reports in Japanese lung cancer patients. Notably, Akkermansia muciniphila and Bifidobacterium breve, which predict immunotherapy response in EGFR-wild-type NSCLC patients, were not identified in this EGFR-mutant cohort. This absence may partly explain why immunotherapy is less effective in EGFR-mutant NSCLC, though further investigation is required.
The association between higher alpha-diversity and milder diarrhea is consistent with broader literature showing that diverse gut microbiota is associated with mucosal resilience and lower susceptibility to intestinal inflammation. This pattern appears to hold regardless of cancer status, as prior Japanese population studies have shown similar trends in healthy individuals without cancer.
Gut microbiota profiling before EGFR-TKI treatment offers a non-invasive biomarker strategy. This first study of gut microbiota in Japanese EGFR-mutant NSCLC patients demonstrated that pretreatment microbiota diversity predicts diarrhea severity and that Ruminococcus abundance predicts EGFR-TKI response. Stool collection is non-invasive and well-tolerated, making microbiota-based biomarker assessment practical in clinical settings.
Key limitations include the small sample size of 21 patients, which limits statistical power and precluded subgroup analyses by EGFR-TKI type. The three EGFR-TKIs used -- osimertinib, afatinib, and erlotinib with ramucirumab -- may have different effects on gut microbiota composition themselves, introducing a confounding variable that could not be separately analyzed in this cohort.
The 16S V1-V2 sequencing approach provides genus-level but not species-level resolution, limiting mechanistic interpretation of Ruminococcus findings. Future studies using metagenomic shotgun sequencing would clarify species-level contributions and their specific metabolic mechanisms. The study also lacked a control group of EGFR-wild-type patients or healthy individuals for direct comparison, and overall survival data were incomplete due to the short study duration.
Future directions include expanding sample size to enable stratified analyses by EGFR-TKI type, incorporating a matched control cohort, applying shotgun metagenomics for deeper taxonomic resolution, and exploring whether microbiota-modulating interventions such as probiotics or dietary modification could improve EGFR-TKI tolerability and efficacy in patients with unfavorable pretreatment microbiome profiles.