Validation of Serum Metabolite Differences in Patients With Gallstones and Gallbladder Cancer

Gallbladder cancer (GBC) is a highly lethal malignant tumour characterized by a low global incidence but a high prevalence in specific populations, particularly in South America. The absence of specific symptoms and validated early detection tools often leads to late diagnosis and poor prognosis. Although altered metabolomic profiles have been described in GBC, and several metabolites have been proposed as potential risk biomarkers, these findings remain largely exploratory and lack consistent validation in independent cohorts, particularly populations with a high incidence of GBC.

Metabolomics provides a comprehensive approach to characterising disease-associated metabolic alterations. Serum metabolites may reflect cancer-related changes in lipid metabolism, bile acid metabolism, amino acid metabolism, and other biological pathways involved in inflammation, carcinogenesis, and tumour progression.

In our previous research, we used the Biocrates platform to obtain serum metabolomics data for 197 Chilean patients, including 86 patients with gallbladder cancer and 111 patients with gallstone disease. The concentrations in serum of more than 1,000 metabolites were processed applying predefined quality-control steps, missing-value filtering, imputation, log2 transformation, and quantile normalisation.

After robust Huber linear regression analysis adjusted for age and sex, the ten metabolites listed below showed the smallest p-values for association with disease status (GBC versus gallstone disease) and a fold-change greater than 1.5. These metabolites have been selected for subsequent validation in the present study:

1. GCA
2. TCDCA
3. GCDCA
4. PG_16:0_20:3
5. PE_44:11
6. PG_16:0_20:4
7. Cer_d18:1/24:1
8. Cer_d18:1/16:0
9. PC_32:0
10. Cer_d18:1/18:0

The primary aim of this study is to validate the differential serum levels of these ten candidate metabolites in an independent Chilean validation cohort comprising 83 patients with GBC and 82 patients with gallstone disease. In addition, we will assess the transferability of these findings to other populations, including a Peruvian cohort with 32 GBC and 26 gallstone disease patients.

Serum metabolite levels will be quantified using Biocrates XL 500 kit and epidemiological and metabolite data will be analysed in R. Global metabolomic profiles will be explored using principal component analysis to identify samples with atypical metabolic patterns, and the final dataset will be analysed following quality control, log2 transformation, and quantile normalisation.

To validate the ten associations previously identified, serum metabolite levels will be compared between patients with GBC and those with gallstone disease in the independent Chilean validation cohort. Statistical analyses will include linear regression using the rlm function from the MASS R package with a Huber tuning constant of k=1.345, adjusted for age (continuous) and sex (binary covariate). P-values will be corrected for multiple testing using the Bonferroni method. The diagnostic performance of metabolites with validated differential levels will be assessed using receiver operating characteristic (ROC) curve analysis, including area under the curve (AUC), sensitivity, and specificity). Similar analyses will be performed to assess the transferability of serum metabolite differences to other patient populations, e.g. Peruvians.