Weighted Net Clinical Benefit (w-NCB)
This method is equivalent to benefit-risk evaluation following a generalized NNT/NNH concept. For the benefit-risk assessment of natalizumab, two different types of data are available: annualised incidence rates (# events / patient year), and incidence proportions (# patients with event / # patients).
The incidence rate for relapses was replaced by the proportion of patients with at least one relapse to allow calculation of weighted NCB using quantitative data. Congenital abnormality was excluded since no quantification was possible and convenience was also removed from the NCB calculation as this was assessed to be a benefit of only secondary importance. For the benefit-risk evaluation, the preference weights were used.
Table 11 Data used for NCB calculation of natalizumab
| Event | Weight | Differences in Proportions: Natalizumab vs. | ||
|---|---|---|---|---|
| Interferon beta-1a | Glatiramer acetate | Placebo | ||
| Disability progression | 5.6 | 0.03 | 0.07 | 0.12 |
| Relapses | 3.9 | 0.17 | 0.21 | 0.26 |
| PML (at time of approval) | 55.9 | -0.001 | -0.001 | -0.001 |
| PML (post-approval) | 55.9 | -0.0015 | -0.0015 | -0.0015 |
| Reactivation of herpes | 6.7 | -0.01 | -0.01 | -0.01 |
| Transaminases elevated | 11.2 | -0.01 | -0.01 | -0.01 |
| Seizure | 5.6 | 0.01 | 0.00 | 0.00 |
| Hypersensitivity Reaction | 1.1 | -0.05 | -0.03 | -0.05 |
| Infusion/injection site reactions | 2.8 | 0.076 | 0.033 | -0.056 |
| Flu-like reactions | 1.1 | 0.21 | 0.00 | 0.00 |
Note: Positive differences indicate a benefit of natalizumab vs. control treatment, negative differences indicate a risk. The numbers given in Table 11 show that, despite the high preference weight, PML will only have little impact on the NCB calculation due to its low frequency. When compared to placebo, PML incidence would need to be at least 0.012 to cancel out the benefit resulting from disability progression, and at least 0.018 to cancel out that from reduction of relapses.
Table 12 Weighted Net Clinical Benefit (w-NCB)
| Submission | After Registration | |
|---|---|---|
| Natalizumab vs. interferon beta-1a | 1.041 | 1.013 |
| Natalizumab vs. glatiramer acetate | 1.036 | 1.008 |
| Natalizumab vs. placebo | 1.239 | 1.211 |
In all comparisons, a positive NCB was achieved (Table 12). As to be expected, the weighted NCB was most pronounced when natalizumab was compared to placebo.
Note:
- This example only uses estimates for NCB. Further information regarding the variability of NCB would be required for a proper benefit-risk assessment.
- The comparison of natalizumab vs. placebo is the most reliable comparison in this setting, because no head-to-head comparisons of natalizumab with the other active compounds are available. As a consequence, comparative data had to be generated by indirect comparisons requiring additional assumptions and resulting in an increased variability of the NCB estimation.
Initially a deterministic sensitivity analysis was employed to obtain a point estimate for benefit-risk and identify the treatment with the highest overall benefit-risk balance. Then a fully stochastic model was used, which involved assigning a probability distribution to each parameter (posterior distribution resulting from a Bayesian analysis of clinical data wherever possible). The distribution of the overall benefit-risk balance for each alternative was then obtained by Monte Carlo simulation, although this means it is not as straightforward to identify the best treatment.