Table 4.
Modeling clinical validity and population impact of variants for a natalizumab PML genetic risk testa.
| Measureb | EUR all (n = 136) | EUR non-MS (n = 121) | EUR MS (n = 15) | EUR MS (n = 15) | AFR all (n = 49) |
|---|---|---|---|---|---|
| Variant IDsc | 1–9 | 1–9 | 1–9 | 1, 3, 4 | 10–19, 9 |
| Presence in PML cases | 33.1% | 30.6% | 53.3% | 33.3% | 53.1% |
| Presence in gnomAD subjects | 8.5% | 8.5% | 8.5% | 1.4% | 7.7% |
| Adverse event frequencyd | 1.3% | 1.3% | 1.3% | 1.3% | 1.3% |
| p-value | 6.23E-16 | 3.30E-12 | 1.01E-05 | 1.23E-06 | 1.71E-16 |
| OR [95% CI] | 5.33 [3.64–7.70] | 4.75 [3.13–7.07] | 12.31 [3.90–39.90] | 36.46 [9.75–117.60] | 13.46 [7.35–24.79] |
| Sensitivity | 32.2% | 29.8% | 51.5% | 27.2% | 51.0% |
| Specificity | 91.8% | 91.8% | 92.1% | 99.0% | 92.8% |
| PPV | 4.9% | 4.6% | 7.9% | 26.1% | 8.6% |
| NPV | 99.0% | 99.0% | 99.3% | 99.0% | 99.3% |
| PAF | 25.9% | 23.3% | 47.0% | 26.2% | 46.9% |
| NNT | 25 | 28 | 14 | 4 | 13 |
| NNG | 297 | 330 | 164 | 294 | 164 |
| NNT/NNG | 0.08494 | 0.08494 | 0.08494 | 0.01353 | 0.07745 |
| MS cases excluded from therapye | 4,672 | 4,672 | 4,672 | 744 | 4,260 |
| PML cases preventedf | 108 | 97 | 196 | 109 | 196 |
Data are reported on an ethnic-specific basis for the combined PML cohorts (Dis and Rep); subsets of the EUR cohort (non-MS, MS all 9 variants, MS top 3 variants) are also reported.
Measure abbreviations reported in Tonk et al. (65): PPV, positive predictive value; NPV, negative predictive; PAF, population attributable fraction; NNT, number needed to treat; NNG, number needed to genotype.
See Table 3 for variants IDs.
Natalizumab manufacturer's maximal reported incidence of PML (13/1,000), accessed on November 9, 2019: www.tysabrihcp.com/en_us/home/efficacy-safety/pml-risk.html.
Estimated number of MS cases excluded from therapy = NNT/NNG × 55,000 JCV-positive MS patients (19).
Estimated number of PML cases prevented = PAF × 418 JCV-positive PML cases (19).