Colorectal cancer, Vitamin D and microbiota: A double-blind Phase II randomized trial (ColoViD) in colorectal cancer patients

Abstract

Background: Several studies suggest a role of gut microbiota in colorectal cancer (CRC) initiation and progression. Vitamin D (vitD) blood levels are also inversely correlated with CRC risk and prognosis. However, these factors' interplay remains unknown.

Methods: 74 CRC patients after standard treatment were randomized to 1-year 2000 IU/day vitD or placebo. Baseline and post-treatment fecal microbiota for shotgun metagenomics sequencing was collected. Coda-lasso and Principal Component Analysis were used to select and summarize treatment-associated taxa and pathways. Associations between vitD and taxa/pathways were investigated with logistic regression. Mediation analysis was performed to study if treatment-associated taxa mediated the effect of supplementation on 25(OH)D levels. Cox proportional-hazards model was used for disease-free survival (DFS).

Results: 60 patients were analyzed. Change in alpha diversity (Shannon: p = 0.77; Simpson: p = 0.63) and post-treatment beta diversity (p = 0.70) were comparable between arms. Post-treatment abundances of 63 taxa and 32 pathways differed between arms. The 63 taxa also mediated the effect of supplementation on 25(OH)D (p = 0.02). There were sex differences in vitD levels, microbiota and pathways. Pathways of essential amino acids' biosynthesis were more abundant in supplemented women. Fusobacterium nucleatum presence at baseline was associated with worse DFS (p = 0.02). Those achieving vitD sufficiency (25(OH)D≥30 ng/ml) had lower post-treatment abundances (p = 0.05). Women were more likely to have F. nucleatum post-treatment (p = 0.02).

Conclusions: VitD supplementation may contribute shaping the gut microbiota and the microbiota may partially mediate the effect of supplementation on 25(OH)D. The observed sex-specific differences highlight the necessity of including sex/gender as a variable in microbiome studies.

Keywords: Colorectal cancer; Gender; Microbiome; Microbiota; Sex; Vitamin D.

Fig. 1

Taxa significantly associated with a. treatment arm b. post-treatment vitamin D sufficiency (25(OH)D ng/ml). For each taxon, results are obtained from a multivariable logistic model including the post-treatment clr-transformed abundance of the taxon as covariate and adjusted for confounders. The bar length indicates the significant beta-coefficient of the taxon (p < 0.05). If positive, the taxon was significantly more abundant in patients a. supplemented with vitamin D b. reaching vitamin D sufficiency at the end of the treatment. If negative, the taxon was significantly more abundant in patients a. in the placebo group b. not reaching vitamin D sufficiency at the end of the treatment. 25(OH)D = 25-hydroxy vitamin D.

Fig. 2

Direct acyclic graph (DAG) of mediation model analyses. The 63 selected taxa (summarized with PC2) as mediator of the effect of vitamin D supplementation (exposure) on post-treatment 25(OH)D levels (outcome). In black, natural direct effect (NDE); in red, natural indirect effect (NDE); in blue, the effect of confounders on the exposure–outcome relationship. p-value obtained from mediation analysis. Significant direct effect of vitamin D supplementation on post-treatment 25(OH)D (p < 0.0001). The 63 taxa significantly mediate the effect of supplementation on post-treatment 25(OH)D (p = 0.02). 25(OH)D = 25-hydroxy vitamin D.

Fig. 3

Pathways significantly associated with a. treatment arm b. post-treatment vitamin D sufficiency (25(OH)D ng/ml). For each pathway, results are obtained from a multivariable logistic model including the post-treatment clr-transformed abundance of the pathway as covariate and adjusted for confounders. The bar length indicates the significant beta-coefficient of the pathway (p < 0.05). If positive, the pathway was significantly more abundant in patients a. supplemented with vitamin D b. reaching vitamin D sufficiency at the end of the treatment. If negative, the pathway was significantly more abundant in patients a. in the placebo group b. not reaching vitamin D sufficiency at the end of the treatment. 25(OH)D = 25-hydroxy vitamin D.