This study included a cohort of 25 patients diagnosed with lung cancer, all of whom underwent bronchoscopy at the Department of Respiratory and Critical Care Medicine, Affiliated Jinhua Hospital, Zhejiang University School of Medicine. The patients were admitted with space-occupying lesions in the lung between January 2024 and July 2024. BALF, serum, and clinical data were collected from each participant. The inclusion criteria for the study were: (1) initial patient visit, (2) subsequent confirmed diagnosis of lung cancer, and (3) performance of bronchoscopy and collection of BALF at our institution, (4) follow-up in our hospital for 4–6 times of chemotherapy. The exclusion criteria were as follows: (1) history of cancer, (2) antibiotic use within the preceding 3 months, and (3) presence of co-infection, immunodeficiency, or liver and kidney dysfunction.

Compete response (CR) and partial response (PR) were defined as chemo-sensitive patients. Stable disease (SD) and progressed disease (PD) were defined as chemo-insensitive patients.

All participants provided informed consent prior to inclusion in the study. The research adhered to the principles outlined in the Declaration of Helsinki and received approval from the Institutional Review Board of Affiliated Jinhua Hospital, Zhejiang University School of Medicine (Ethics Approval Number: AL-JHYY202437).

BALF was collected from affected pulmonary lobes using sterile tubes under standardized protocols to minimize oral contamination. After centrifugation (15,000 g, 30 min, 4°C), supernatants and cell pellets were stored at −80°C for metabolomics and 16S rDNA sequencing, respectively.

Genomic DNA was extracted from BALF pellets, and the V3–V4 regions of the 16S rRNA gene were amplified using standard primers (341F/806R). Libraries were prepared with dual-indexed barcodes and sequenced on the Illumina MiSeq platform (2 × 250 bp).

BALF metabolites were analyzed using a UHPLC-Q Exactive HF-X system (Thermo Scientific) with an ACQUITY UPLC HSS T3 column (2.1 × 100 mm, 1.8 μm). Mobile phases consisted of (A) 0.1% formic acid in water and (B) acetonitrile. Gradient elution: 0–12 min, 5–95% B; 12–14 min, 95% B; 14–14.1 min, 95–5% B; 14.1–16 min, 5% B. Mass spectrometry operated in positive/negative ion modes with spray voltage ±3.8 kV and resolution 70,000.

Bile acids were quantified utilizing liquid chromatography-tandem mass spectrometry, which included the measurement of five free bile acids—cholic acid (CA), chenodeoxycholic acid (CDCA), deoxycholic acid (DCA), lithocholic acid (LCA), and ursodeoxycholic acid (UDCA)—alongside 10 conjugated bile acids: glycocholic acid (GCA), glycochenodeoxycholic acid (GCDCA), glycodeoxycholic acid (GDCA), glycolithocholic acid (GLCA), glycoursodeoxycholic acid (GUDCA), taurocholic acid (TCA), taurochenodeoxycholic acid (TCDCA), taurodeoxycholic acid (TDCA), taurolithocholic acid (TLCA), and tauroursodeoxycholic acid (TUDCA). This analytical procedure was performed by Dian Diagnostic Technology Co., Ltd.

To compare the abundances of individual taxa between two groups, the STAMP software was employed, while LEfSe was utilized for the quantitative analysis of biomarkers. Differences in microbial communities were evaluated using ANOSIM and ADONIS, based on Bray–Curtis dissimilarity matrices. Additional analyses included principal component analysis (PCA) and multivariate statistical assessments with SIMCA 14.1, Pearson correlation analysis via CytoScape 3.5.1, and KEGG pathway analysis conducted using R 3.5.1. Statistical significance was denoted by p-values, with * indicating p < 0.05, ** indicating p < 0.01, and *** indicating p < 0.001.

In this study, we included 25 patients diagnosed with lung cancer via bronchoscopy and subsequently treated with chemotherapy at our hospital. These patients were categorized into two groups: a chemotherapy-sensitive group comprising 15 individuals and a chemotherapy-insensitive group consisting of 10 individuals. BALF was collected from all participants, and both microbiological and metabolomic analyses were conducted. The baseline characteristics of the two patient groups in the study cohort were detailed in Table 1. Our analysis revealed no significant differences between the groups in terms of age, sex, pathological type, liver function (TBA, ALT, AST, and TBIL), or tumor markers (CEA, CA199, CA125, SCC, CK19, NSE, and ProGRP).

Metabolites are crucial in the onset, progression, and therapeutic response of tumors, particularly in lung cancer. Alterations in alveolar metabolites can directly influence the efficacy of chemotherapy. To elucidate the relationship between lung metabolites and chemotherapy sensitivity in lung cancer, we categorized the enrolled patients into two cohorts. Comparative analyses indicated no statistically significant differences between the groups concerning gender, age, histological type, and laboratory results (p > 0.05). principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA) (Figure 1A; Supplementary Figure 1A), along with their respective permutation tests (Supplementary Figure 1B), identified a distinct set of differential metabolites associated with the two groups.

We utilized ultra-high-performance liquid chromatography coupled with quadrupole time-of-flight mass spectrometry (UHPLC-Q-TOF MS) to perform high-resolution untargeted metabolomics analysis for metabolite identification. This methodology facilitated the detection of 1,629 metabolites in the positive ion mode and negative ion mode. To discern differential metabolites in the BALF between the two groups of lung cancer patients, we identified significant differences in metabolite levels based on the variable importance in projection (VIP) score, employing a significance threshold of p < 0.05 and a fold change greater than 1.5. This analysis resulted in the identification of 92 metabolites, of which 32 were significantly elevated and 60 were significantly decreased in the chemotherapy-sensitive group (Figure 1B and Table 2). In the chemotherapy-sensitive group, trans-urocanate had the highest increase, while Phenylalanylphenylalanine had the greatest decrease among the abundant metabolites (Figure 1C).

We used STITCH to build a metabolic network of correlated differential metabolites to examine their interactions. Figure 1D shows that all these metabolites are interconnected, either directly or indirectly. Strong interactions were observed between the linoleic acid amide, petroselaidic acid (C18-1 T), oleamide, and oleoyl ethanolamide (all interaction scores >0.9).

KEGG pathway enrichment analysis was performed by Fisher’s exact test to identify metabolic and signal transduction pathways significantly associated with chemo sensitivity in lung cancer patients. The results showed that ABC transporters, biosynthesis of unsaturated fatty acids, glutathione metabolism, and histidine metabolism pathways were significantly enriched (Figures 1E,F).

Our GSEA analysis of differential metabolites revealed significant enrichment in metabolic pathways, biosynthesis of amino acids, ABC transporters, and bile secretion pathways (Supplementary Figures 1C–F). Furthermore, AUC (area under the ROC curve) analysis revealed several metabolites with high sensitivity and specificity in differentiating lung cancer patients with chemotherapy sensitive from those with insensitive, including dodecanamide (AUC = 0.927), N-methyl-aspartic acid (AUC = 0.920) (Figure 1G).

In the present microbiome investigation, we successfully performed PCR amplification of the 16S rDNA gene, specifically targeting the V3–V4 regions, across all collected samples. The Venn diagram depicted in Figure 2A, derived from the ASV table data, visually represents the quantity of shared and unique ASVs for each group. The analysis revealed that the two groups shared 1,116 ASVs, with the sensitive group possessing 3,741 ASVs and the insensitive group possessing 2,719 ASVs.

Three diversity indices—Chao 1, Shannon, and observed species—were employed to assess α-diversity between the groups. The analysis revealed no significant differences, indicating that there was no notable alteration in lung microbial community diversity between the two groups (Figure 2B). To further examine the similarities in bacterial communities between the groups, a principal coordinates analysis (PCoA) plot was generated using unweighted UniFrac distances at the operational taxonomic unit (OTU) level. The results demonstrated no discernible separation between the groups, suggesting that the primary composition of the lung microbiome in the cohort with lung cancer and distant metastasis remained largely unchanged (Figure 2C).

A subsequent analysis of the compositional disparities in the pulmonary microbiota between the two patient cohorts demonstrated that, although numerous taxa were shared, their relative abundances exhibited significant differences. Notably, substantial variations were observed in the proportions of Stenotrophmonas, Streptococcus, Haemophilus, Pseudomonas, Neisseria, Veillonella, Tropheryma, and Staphylococcus between the two groups, as illustrated in Figures 2D,E. Moreover, a t-test highlighted microecological variations between the two groups, showing a notable increase in the Caulobacter (p = 0.003), CL500-29_marine_group (p = 0.002), [Eubacterium]_eligens_group (p = 0.0046) and decrease in the Acinetobacter (p = 0.031), Abiotrophia (p = 0.010) genus within the alveolar lavage fluid of sensitive patients (Table 3).

The predicted functions of bacteria associated with lung cancer were analyzed utilizing PICRUSt2. Ten key functions were determined to be most significant for lung-associated bacteria, as illustrated in Figure 2F. Among these, ABC transport system-related pathways were significantly clustered (see Figure 3).

To examine the phenotypic modifications that may be elicited by changes in the host’s microbial community structure, we performed an association analysis between the metabolome and the microbiome. Initially, a correlation analysis was performed, and a heatmap was generated using the Pearson correlation coefficient to investigate the significantly different genera identified through 16S rDNA analysis, as well as the significantly different metabolites identified through metabolomics analysis. This study sought to evaluate the degree of association between species diversity and metabolites in environmental samples. The findings revealed a positive correlation between CL500-29_marine_group and Pentadecanolide (r = 0.412; p = 0.041), both of which were significantly elevated in the alveolar lavage fluid of patients sensitive to chemotherapy (Figures 4A,B).

A chord plot and a Sankey diagram was developed using the correlation coefficients between differential bacterial genera and differential metabolites, providing a visual representation of the associations between key metabolites and key bacterial genera (Figures 4C,D).

The findings indicate a correlation between the cholic acid concentration in alveolar lavage fluid and the chemosensitivity of lung cancer. This association is particularly noteworthy given that bile acids are commonly present in the digestive system. Consequently, we conducted a targeted metabolomic analysis of bile acids in pre-chemotherapy blood samples obtained from these patients.

The results revealed that the concentrations of four specific bile acids, chenodeoxycholic acid (CDCA), cholic acid (CA), deoxycholic acid (DCA), and ursodeoxycholic acid (UDCA) were notably higher in the chemotherapy sensitive lung cancer patients, with p-values of 0.007, 0.005, 0.035, and 0.017, respectively (Figure 5A). No significant difference was found in other bile acids (Supplementary Figure 2). We conducted a ROC curve analysis on the four bile acids to assess their diagnostic value for lung cancer, finding AUC values of 0.830 (CA), 0.813 (CDCA, UDCA), and 0.633 (DCA) (Figure 5B).