Table 1 Validation of the LV mass polygenic predictor (PRS).

From: The polygenic architecture of left ventricular mass mirrors the clinical epidemiology

Data source

Independent variable

Additional covariatea

OR (95% CI)b

P-value

Derivation setc

LV Mass measured

None

2.1 (2.0–2.3)

<2 × 10−16

LV Mass PRS

None

2.0 (1.9–2.1)

<2 × 10−16

LV Mass PRS

LV Mass measured

1.1 (0.99–1.31)

0.06

LV Mass PRS

Height

2.0 (1.9–2.2)

<2 × 10−16

LV Mass PRS

BMI

1.9 (1.8–2.1)

<2 × 10−16

LV Mass permuted PRS

None

0.99 (0.94–1.04)

0.66

Validation set 1: BioVUd

LV Mass PRS

None

1.08 (1.02–1.13)

0.002

Validation set 2: eMERGEe

LV Mass PRS

None

1.10 (1.06–1.14)

4.1 × 10−7

  1. aAdditional covariate added to the logistic regression model.
  2. bAnalyses are based on a logistic regression model using the PheWAS phenotype cardiomegaly as the dependent variable and specified independent variable. All independent variables were set to have a standard deviation of 1. All models were additionally adjusted for age, sex and 5 principal components.
  3. cThe data set used to develop the PRS. There were 2929 cases and 3416 controls for cardiomegaly.
  4. dCases (n = 2,624) and controls (n = 16,682) for the cardiomegaly phenotype that were not part of the derivation set.
  5. eCases (n = 5,982) and controls (n = 21,198) for the cardiomegaly phenotype that were not part of the derivation set.
Back to article page