TABLE 4b.
Model estimates, predictive diagnostics, and relative contribution of PheRS in the presence of covariates in MGI.
| Model |
||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Covariates only | Covariates + risk factors | PRS only | PRS + covariates + risk factors | PheRS only | PheRS + covariates + risk factors | PheRS + PRS only | PheRS, PRS, covariates + risk factors | |||
| 0 | PheRS | Odds Ratio | - | - | - | - | 7.58 | 7.70 | 7.57 | 7.69 |
| 95% CI | - | - | - | - | 6.01, 9.75 | 6.08, 9.97 | 6.00, 9.76 | 6.06, 9.97 | ||
| PRS | Odds Ratio | - | - | 1.59 | 1.59 | - | - | 1.51 | 1.47 | |
| 95% CI | - | - | 1.14, 2.24 | 1.13, 2.25 | - | - | 0.94, 2.43 | 0.90, 2.39 | ||
| Diagnostics | AUC | 0.557 | 0.609 | 0.555 | 0.619 | 0.852 | 0.885 | 0.864 | 0.887 | |
| AUC 95% CI | 0.520, 0.594 | 0.573, 0.645 | 0.517, 0.593 | 0.583, 0.655 | 0.827, 0.876 | 0.862, 0.908 | 0.838, 0.891 | 0.865, 0.910 | ||
| HL Stat (P-value) | 3.59 (8.92E-01) | 8.28 (4.06E-01) | 12.59 (1.27E-01) | 4.09 (8.49E-01) | 0.78 (1.00E+00) | 12.63 (1.25E-01) | 9.97 (2.67E-01) | 16.05 (4.17E-02) | ||
| Brier Score | 0.22870 | 0.22330 | 0.22890 | 0.22150 | 0.11600 | 0.11290 | 0.11630 | 0.1130 | ||
| 1 | PheRS | Odds Ratio | - | - | - | - | 3.68 | 3.74 | 3.69 | 3.75 |
| 95% CI | - | - | - | - | 2.72, 5.21 | 2.74, 5.35 | 2.72, 5.22 | 2.75, 5.36 | ||
| PRS | Odds Ratio | - | - | 1.11 | 1.19 | - | - | 1.18 | 1.23 | |
| 95% CI | - | - | 0.70, 1.77 | 0.74, 1.92 | - | - | 0.71, 1.95 | 0.73, 2.09 | ||
| Diagnostics | AUC | 0.575 | 0.604 | 0.524 | 0.606 | 0.717 | 0.755 | 0.730 | 0.758 | |
| AUC 95% CI | 0.525, 0.624 | 0.556, 0.652 | 0.473, 0.575 | 0.557, 0.654 | 0.678, 0.755 | 0.713, 0.798 | 0.686, 0.774 | 0.716, 0.800 | ||
| HL Stat (P-value) | 9.25 (3.22E-01) | 11.34 (1.83E-01) | 18.77 (1.62E-02) | 14.04 (8.07E-02) | 2.16 (9.76E-01) | 6.95 (5.43E-01) | 10.07 (2.60E-01) | 5.48 (7.06E-01) | ||
| Brier Score | 0.22060 | 0.21820 | 0.22430 | 0.21790 | 0.18100 | 0.17590 | 0.18070 | 0.17560 | ||
| 2 | PheRS | Odds Ratio | - | - | - | - | 3.62 | 3.68 | 3.72 | 3.79 |
| 95% CI | - | - | - | - | 2.63, 5.17 | 2.66, 5.29 | 2.70, 5.35 | 2.73, 5.50 | ||
| PRS | Odds Ratio | - | - | 1.21 | 1.32 | - | - | 1.55 | 1.67 | |
| 95% CI | - | - | 0.74, 1.97 | 0.80, 2.20 | - | - | 0.91, 2.65 | 0.96, 2.91 | ||
| Diagnostics | AUC | 0.576 | 0.609 | 0.530 | 0.611 | 0.704 | 0.744 | 0.729 | 0.750 | |
| AUC 95% CI | 0.523, 0.628 | 0.558, 0.661 | 0.476, 0.584 | 0.560, 0.663 | 0.663, 0.746 | 0.699, 0.789 | 0.682, 0.775 | 0.705, 0.795 | ||
| HL Stat (P-value) | 13.21 (1.05E-01) | 9.96 (2.68E-01) | 14.98 (5.96E-01) | 11.59 (1.70E-01) | 4.31 (8.28E-01) | 17.37 (2.65E-02) | 12.29 (1.39E-01) | 8.32 (4.03E-01) | ||
| Brier Score | 0.21980 | 0.21660 | 0.22370 | 0.21610 | 0.18720 | 0.18140 | 0.18570 | 0.17940 | ||
| 5 | PheRS | Odds Ratio | - | - | - | - | 5.62 | 5.65 | 5.60 | 5.59 |
| 95% CI | - | - | - | - | 3.42, 10.60 | 3.43, 10.70 | 3.41, 10.60 | 3.39, 10.50 | ||
| PRS | Odds Ratio | - | - | 1.52 | 1.70 | - | - | 1.61 | 1.81 | |
| 95% CI | - | - | 0.87, 2.68 | 0.95, 3.09 | - | - | 0.85, 3.06 | 0.93, 3.57 | ||
| Diagnostics | AUC | 0.575 | 0.621 | 0.552 | 0.629 | 0.697 | 0.766 | 0.728 | 0.774 | |
| AUC 95% CI | 0.514, 0.636 | 0.562, 0.679 | 0.488, 0.615 | 0.571, 0.687 | 0.655, 0.739 | 0.715, 0.816 | 0.672, 0.784 | 0.725, 0.824 | ||
| HL Stat (P-value) | 8.30 (4.04E-01) | 7.45 (4.89E-01) | 6.79 (5.60E-01) | 9.40 (3.10E-01) | 1.64 (9.90E-01) | 8.00 (4.34E-01) | 12.84 (1.18E-01) | 13.81 (8.68E-01) | ||
| Brier Score | 0.22020 | 0.21500 | 0.22230 | 0.21350 | 0.17330 | 0.16790 | 0.17180 | 0.16670 | ||
Abbreviations: AUC, area under the receiver-operator characteristics curve; CI, confidence interval; HL, Hosmer-Lemeshow Chi-Square test; PheRS, phenotype risk score; PRS, polygenic risk score.
Notes: Variable estimates come from a logistic GLM model. PheRS constructed using multivariate betas obtained in MGI and applied to time-restricted, matched MGI data. Diagnostic results are obtained on using the fitted values from the corresponding model. The covariates used are sex, age at time threshold, genotyping batch, and the first four principal components from the genotype data. The risk factors are BMI (continuous), drinker (ever/never), and smoker (ever/never).