As an exploratory analysis, we estimated PPS for biomarkers of lipid metabolism.29 (link) The PPS estimates the sum of additive genetic effects across all alleles that affect the biomarkers of lipid metabolism at the patient level. We used the PPS to explore a potential genetic link between lipid metabolism, ALS, and survival time by assessing (1) how much of the variance in biomarker levels at diagnosis can be explained by genetic profile scores and (2) whether the genetic profile score itself is associated with overall survival time. Because PPS does not change over time,30 (link) a statistical association between the genetic profile score and survival may be evidence of abnormal lipid levels caused by genetic variation or hold potential for therapeutic interventions.30 (link) Moreover, their time invariance allowed us to estimate the link between the genetic profile score and overall survival time, defined as time between symptom onset and death.
For all individuals who were enrolled in both our population-based registry and our latest genome-wide association study (GWAS),5 (link) we calculated the PPS. PPS was based on summary statistics from a GWAS on biomarker levels of lipid metabolism in the UK Biobank.31 For each single-nucleotide polymorphism, we calculated a weight for each biomarker using the summary-BavesR module in the Genome-Wide Complex Trait Bayesian analysis toolkit (default parameters)29 (link) and a linkage-disequilibrium matrix originating from 50,000 unrelated individuals of inferred European ancestries included in the UK Biobank. Because the genotype data originated from several different cohorts in the ALS GWAS, we scaled the PPS per GWAS cohort to a mean of zero and a standard deviation of 1. Linear regression models were used to calculate how much of the variance in the biomarker level was explained by their PPS (expressed as adjusted R2); 95% confidence intervals were obtained by means of bootstrapping. Simple univariable Cox models for overall survival time (i.e., from onset to death) were used to estimate HRs.
For all individuals who were enrolled in both our population-based registry and our latest genome-wide association study (GWAS),5 (link) we calculated the PPS. PPS was based on summary statistics from a GWAS on biomarker levels of lipid metabolism in the UK Biobank.31 For each single-nucleotide polymorphism, we calculated a weight for each biomarker using the summary-BavesR module in the Genome-Wide Complex Trait Bayesian analysis toolkit (default parameters)29 (link) and a linkage-disequilibrium matrix originating from 50,000 unrelated individuals of inferred European ancestries included in the UK Biobank. Because the genotype data originated from several different cohorts in the ALS GWAS, we scaled the PPS per GWAS cohort to a mean of zero and a standard deviation of 1. Linear regression models were used to calculate how much of the variance in the biomarker level was explained by their PPS (expressed as adjusted R2); 95% confidence intervals were obtained by means of bootstrapping. Simple univariable Cox models for overall survival time (i.e., from onset to death) were used to estimate HRs.
Free full text: Click here
