TL;DR
This study leverages transcriptome-wide association studies (TWAS) to identify novel genetic loci and candidate genes associated with cutaneous squamous cell carcinoma (cSCC). By imputing gene expression levels from genotype data, the researchers discovered 15 genes in 9 loci linked to cSCC risk, including 8 genes in 5 previously unknown susceptibility regions, offering new insights into the genetic underpinnings of this common skin cancer.
What Problem Does This Paper Address?
Genome-wide association studies (GWAS) have successfully identified genetic loci associated with various diseases, including cutaneous squamous cell carcinoma (cSCC). However, GWAS typically pinpoint regions containing multiple genes, making it challenging to identify the specific causal gene responsible for the observed association. This ambiguity hinders our understanding of the biological mechanisms driving disease risk. Transcriptome-wide association studies (TWAS) offer a complementary approach by directly linking genetic variation to gene expression and then associating that expression with disease risk, aiming to pinpoint causal genes more effectively. This paper addresses the need to move beyond broad genomic regions to specific genes and their functional roles in cSCC pathogenesis.
Methods
The core methodology employed is Transcriptome-Wide Association Studies (TWAS), specifically using the prediXcan method. PrediXcan trains linear regression models to impute gene expression levels in specific tissues based on genotype data from a reference panel (GTEx Consortium). These imputation models are then applied to genotype data from large case-control cohorts to predict gene expression levels for each individual. These imputed expression levels are subsequently tested for association with cSCC risk. The study focused on four "disease-relevant" tissue types: sun-exposed skin, non-sun-exposed skin, whole blood, and lymphocyte cell lines (LCLs). A two-phase discovery-validation approach was used, with the Kaiser Permanente Genetic Epidemiology Research in Adult Health and Aging (GERA) cohort serving as the discovery set and the 23andMe research participant dataset as the validation set. Genes were considered significant if they met a false discovery rate (FDR) < 10% in discovery and Bonferroni significance in validation.
Datasets
- Kaiser Permanente Genetic Epidemiology Research in Adult Health and Aging (GERA) cohort:
- Discovery set: 6,891 cSCC cases and 54,566 non-cSCC controls.
- Genotyping: Custom Affymetrix Axiom array, imputed to 1000 Genomes Project SNPs.
- 23andMe research participant dataset:
- Validation set: 25,558 self-reported cSCC cases and 673,788 self-reported non-cSCC controls.
- Genotyping: Four 23andMe genotyping chips, imputed to 1000 Genomes Project SNPs.
- Genotype-Tissue Expression (GTEx) Consortium V6p dataset:
- Used as a reference panel for training prediXcan imputation models.
Key Insights and Ideas
- TWAS as a powerful complement to GWAS: The study demonstrates that TWAS can identify novel genetic associations that might be missed by traditional GWAS, particularly for genes with smaller effect sizes or complex genetic architectures.
- Pinpointing candidate causal genes: By linking genetic variation to gene expression, TWAS provides a more direct path to identifying specific genes that may be driving disease risk, moving beyond broad genomic regions.
- Biological plausibility of findings: Many of the identified genes have known roles in cellular processes relevant to skin cancer development, such as immune response, DNA repair, and cell signaling, lending biological support to the statistical associations.
- Challenges in imputation quality: The study highlights that imputation quality can vary significantly across tissues and genes, impacting the reliability of TWAS findings. This is particularly relevant for genes involved in complex pathways like pigmentation.
- Interplay between genetic and environmental factors: The discussion touches upon how gene expression might interact with environmental exposures like UV radiation to influence cSCC risk.
Key Results
- 19 loci and 33 genes identified in discovery: In the GERA cohort, 33 genes in 19 loci showed imputed expression levels associated with cSCC at an FDR < 10%.
- 15 genes in 9 loci validated: In the 23andMe dataset, 15 of these genes were validated at Bonferroni significance.
- 5 novel susceptibility loci: Eight of the validated genes were located in five previously unidentified cSCC susceptibility loci.
- Examples include CTSS, HORMAD1, GOLPH3L, ANXA9 at 1q21; CASP8 at 2q33; AHI1 at 6q23; HAL at 12q23; and ORMDL3 at 17q21.
- 7 genes in 4 previously associated loci: Seven validated genes were found in loci previously identified by GWAS.
- Examples include HLA-DOB, SKIV2L, HLA-DRB5 at 6p21; HERC2 at 15q13; CDK10, FANCA at 16q24; and FAM83C at 20q11.
- Direction of association provides mechanistic clues: For many genes, the direction of the association (positive or negative) between imputed expression and cSCC risk aligns with their known biological functions, suggesting potential roles in disease pathogenesis.
Limitations and Future Work
- Imputation quality: The authors acknowledge that imputation quality can be a significant limitation, especially for genes in highly polymorphic regions (like HLA) or for genes whose expression is not well captured by the available reference data. This can make it difficult to definitively identify causal genes.
- Bulk tissue data: The skin tissue expression data used for imputation were from bulk tissue, not specific cell types (e.g., keratinocytes), which can introduce noise and obscure cell-type-specific effects.
- Statistical associations vs. causality: TWAS, like GWAS, identifies statistical associations. Experimental validation is crucial to confirm differential gene expression in cSCC cases and controls and to elucidate the precise causal mechanisms.
- Distinguishing independent associations: At loci with multiple associated genes (e.g., 1q21), it can be challenging to disentangle independent signals from correlated ones, requiring further investigation.
- Future work: The authors emphasize the need for experimental studies to validate findings and explore the biological mechanisms. They also highlight the importance of improving gene expression imputation methods to enhance accuracy and power for future TWAS. Further research is needed to understand the complex interplay of genetic and environmental factors in cSCC.
Summary updated