Disentangling the aetiological pathways between body mass index and site-specific cancer risk using tissue-partitioned Mendelian randomisation

• Objective: This study used tissue-partitioned Mendelian randomization (MR) to separate the genetic effects of BMI into brain-mediated (appetite) and adipose-mediated (metabolic) pathways to assess their independent causal impact on seven site-specific cancer risks.
• Key Finding (Lung Cancer): The brain-mediated BMI variants were identified as the predominant causal driver of lung cancer risk (OR: 1.17), an effect which was shown to be highly correlated with an increase in cigarettes per day, suggesting a shared genetic mechanism acting via addictive behaviors.
• Key Finding (Colorectal Cancer): A non-significant trend suggested that the adipose-mediated BMI variants were more strongly associated with colorectal cancer risk, supporting the role of metabolic consequences of adiposity in this disease.

PubMed: 36434155 DOI: 10.1038/s41416-022-02060-6 Overview generated by: Gemini 2.5 Flash, 28/11/2025

Key Findings: Tissue-Specific BMI Effects on Cancer

This study applied tissue-partitioned Mendelian randomization (MR) to separate the genetic effects of BMI mediated by subcutaneous adipose tissue (metabolic) from those mediated by brain tissue (appetite/central regulation) and assessed their differential causal impact on the risk of seven site-specific cancers.

• Brain-Mediated BMI Drives Lung Cancer Risk: The most distinct finding was that the brain-tissue-derived BMI variants were the predominant driver of the genetically predicted causal effect of BMI on lung cancer risk (OR: 1.17; 95% CI: 1.01-1.36).
  • This effect was strongly supported by a parallel finding: the brain-mediated BMI variants were also robustly associated with an increased number of cigarettes per day (Beta = 0.44; 95% CI: 0.26-0.61). This suggests that the genetic pathway influencing BMI via appetite/central regulation may increase lung cancer risk primarily by promoting addictive behaviors (smoking).
• Adipose-Mediated BMI Drives Colorectal Cancer Risk (Marginal Evidence): The adipose-tissue-derived BMI variants showed a stronger, though non-significant at the conventional level, association with colorectal cancer risk (OR: 1.07; 95% CI: 0.99-1.15). This suggests that the metabolic consequences of excess adipose tissue are the more relevant pathway for this cancer type.
• No Distinct Partitioned Effects on Other Cancers: For the remaining five site-specific cancers investigated (breast, prostate, kidney, ovarian, and endometrial), the causal effects of the brain-mediated and adipose-mediated BMI partitions were statistically indistinguishable. This suggests that either the downstream cancer mechanisms are shared, or the current tissue partitioning is insufficient to separate the relevant pathways for these cancers.

Study Design and Methods

Study Design

This was a two-sample tissue-partitioned Mendelian randomization (MR) study. The methodology leveraged tissue-specific gene expression data to define two distinct sets of genetic instrumental variables (IVs) for BMI, allowing for the comparison of their independent causal effects on cancer risk.

Data and Genetic Instrument Selection

1. Exposure GWAS (BMI): Summary statistics were used from a large-scale BMI GWAS (GIANT consortium, \(n \sim 322,000\)).
2. Tissue-Specific Prioritization: The study used a previously developed method to partition BMI IVs based on colocalization with gene expression (eQTLs) in two relevant tissues:
   • Subcutaneous Adipose Tissue: Used to proxy metabolic/peripheral mechanisms.
   • Brain Tissue: Used to proxy appetite/central regulation mechanisms.
3. Instrument Sets: This resulted in two, largely non-overlapping sets of IVs: the adipose-mediated set and the brain-mediated set.
4. Outcome GWAS Data (Cancers): Outcome data were sourced from large international consortia GWAS for seven site-specific cancers: lung, colorectal, breast, prostate, kidney, ovarian, and endometrial cancers. Data for Cigarettes per day (CPD) were also used as an intermediate outcome to test the smoking pathway hypothesis.

Statistical Analysis

• MR Methodology: A Multivariable Mendelian Randomization (MVMR) approach was applied. This method simultaneously models the effects of the brain-mediated and adipose-mediated BMI partitions, which helps to account for the correlation between the two instrumental variable sets, thereby providing the independent causal effect of each pathway.
• Primary MR Method: The MVMR estimates were derived and the results were reported as Odds Ratios (ORs) per one standard deviation (SD) increase in BMI for each pathway.
• Sensitivity Analyses: Standard MR sensitivity analyses (e.g., MR-Egger) were applied to the aggregated BMI instrument set to ensure overall robustness.

Conclusions and Recommendations

The study successfully demonstrates that the complex genetic effects of BMI can be dissected into biologically meaningful pathways that have distinct, site-specific causal consequences for cancer risk.

• Specific Aetiology for Lung Cancer: The strong evidence linking the brain-mediated/appetite pathway to lung cancer risk, mediated by smoking behavior, provides a specific mechanistic target for prevention. Interventions should focus on the behavioral and neurobiological aspects of appetite and addiction in individuals with high genetic risk for this pathway.
• Colorectal Cancer Focus: The trend toward the adipose-mediated pathway driving colorectal cancer risk supports the traditional hypothesis that this cancer type is linked to the metabolic and inflammatory consequences of excess adiposity.
• Future Directions: The authors suggest that future studies should investigate whether the differential pathways identified for T2DM (previously shown to be brain-mediated) and CAD (shown to be similarly affected by both pathways) could also inform cancer risk mechanisms, further linking cardiometabolic and oncologic outcomes.