PRSWeb

University of Michigan logo

University of Michigan Center for Precision Health Data Science

This webpage provides detailed PheWAS results for three skin cancer subtypes: basal cell carcinoma, squamous cell carcinoma, and melanoma. PRS results are available for PRS constructed using different weights (from the latest GWAS or the NHGRI-EBI GWAS catalog). These results are provided for Michigan Genomics Initiative (MGI), a longitudinal biorepository effort within Michigan Medicine, and for the population-based UK Biobank dataset.

For more information, see our associated publication Fritsche et al. 2019, PLOS Genetics.

PRSweb update with additional cancer traits in development and available at http://prsweb.sph.umich.edu.

>
>
Figure 1: Basal Cell Carcinoma PRS (172.21)circulatory systemcongenital anomaliesdermatologicdigestiveendocrine/metabolicgenitourinaryhematopoieticinfectious diseasesinjuries & poisoningsmental disordersmusculoskeletalneoplasmsneurologicalpregnancy complicationsrespiratorysense organssymptoms-log10 p-value0481216204060Disorder of skin and subcutaneous tissue NOSDegenerative skin conditions and other dermatosesActinic keratosisSeborrheic keratosisOther specified diseases of sebaceous glandsDermatitis due to solar radiationChronic dermatitis due to solar radiationSkin cancerMelanomas of skin, dx or hxMelanomas of skinOther non-epithelial cancer of skinBasal cell carcinomaSquamous cell carcinomaCarcinoma in situ of skinNeoplasm of uncertain behavior of skin
Figure 2: Basal Cell Carcinoma PRS (172.21) (exclusion)circulatory systemcongenital anomaliesdermatologicdigestiveendocrine/metabolicgenitourinaryhematopoieticinfectious diseasesinjuries & poisoningsmental disordersmusculoskeletalneoplasmsneurologicalpregnancy complicationsrespiratorysense organssymptoms-log10 p-value0481216204060

Figure 3: Basal cell carcinoma

12345continuousQ1 vs Q2Q1 vs Q3Q1 vs Q4Odds Ratio (95% CI)

LEGENDS

Figure 1: PheWAS results
P-values correspond to associations between the polygenic risk score (PRS) and the electronic health record-derived phenotype obtained using Firth’s logistic regression and also adjusting for age, gender, genotyping array, and the first four genotype principal components.
Upward (downward)-pointing triangles indicate a positive (negative) association.
Additional details are available by hovering the cursor over a particular triangle and by clicking the triangle. The horizontal dashed line indicates phenome-wide significance. Results are color-coded by disease category.

Figure 2: Exclusion PheWAS results
Results from a PheWAS performed using only subjects who never had a skin cancer diagnosis. The results are obtained as in Figure 1 but using the reduced dataset.

Figure 3: Associations between PRS and Selected Phenotype
This figure shows results for the selected phenotype in Figure 1.
This figure provides the beta estimate for the adjusted association between the PRS and the selected phenotype from Firth-corrected logistic regression. The corresponding confidence intervals are also shown. Results are presented for models using either a continuous or a categorical version of the PRS. Q1 through Q4 represent the four quartiles of the PRS.

CONTACT

Site created by Peter VandeHaar and Lars Fritsche, last updated on 2019 December 06.

Contributors: Lars Fritsche, Lauren J Beesley, and Bhramar Mukherjee

Contact: Lars Fritsche (larsf@umich.edu) and Bhramar Mukherjee (bhramar@umich.edu), 1415 Washington Heights, Ann Arbor MI, 48109