In this post, I’m following up on the key points on the real-world evidence (RWE) data collection in Germany with an IQWiG report on the use of RWE in dossiers.
In an early benefit assessment, the G-BA determines the additional benefit of an intervention compared with the appropriate comparator therapy (ACT). For orphan drugs or products with accelerated approval, the data available at the time of early benefit assessment is often immature or insufficient to derive meaningful conclusions. The studies are mostly too short or non-comparative, which makes it impossible to derive an additional benefit.
Therefore, in 2020, the IQWiG developed criteria on how to supplement the trial data with real-world evidence (German: “Versorgungsnahe Daten”), such as registry data. If the RWE is of high quality, the IQWiG and G-BA can use it to determine the additional benefit of an intervention outside of a clinical trial.
Therefore, the IQWiG report from 2020 includes detailed recommendations for manufacturers and registries.
IQWiG report after 2 years
Two years after the IQWiG shared these RWE recommendations, they assessed how manufacturers incorporated RWE in their dossiers. Unfortunately, they concluded that “manufacturers don’t interpret the detailed requirements on RWE collection how we wish they did”.
Overall, the problems often were missing key information and differences between patient populations. It was not possible to conduct a meaningful benefit assessment with the data. Manufacturers often did identify the problems themselves but did not draw the necessary consequences.
As examples, the IQWiG report specifically describes the problems seen in the four benefit assessments for amivantamab, nivolumab, dostarlimab and lanadelumab.
Amivantamab – Missing data on patient selection and risk adjustments
Amivantamab was assessed as a monotherapy for the treatment of adults with advanced non-small-cell lung cancer. The manufacturer used data from two registries to derive an additional benefit.
However, the registries were missing information on important patient characteristics, such as health status, the severity of disease, and prior treatments. In addition, there were no data on key endpoints.
The manufacturer was aware of these data gaps and identified these as relevant confounding factors, but nonetheless aimed to use the data to claim an additional benefit.
Nivolumab – Relevant risk differences in the population
Nivolumab in combination with ipilimumab was assessed in adults with colorectal cancer. The manufacturer submitted routine data from the US health database Flatiron.
The key problem of the data source is the risk difference. The mortality in the study was 10% higher in patients from the US compared with European patients. Again, the manufacturer identified this confounding factor, but did not account for it in the analysis.
In addition, the registration trial for nivolumab only included patients without any side effects to prior treatments or conspicuous laboratory values. The Flatiron data did not include this level of information though. Thus, it was not possible to select a patient population for the comparator arm matching the nivolumab trial population. Therefore, there was a relevant risk difference between both populations, which the manufacturer did not consider.
Dostarlimab – Biomarker not included in the data for the comparator therapy
Dostarlimab was assessed for the treatment of patients with endometrial cancer with tumours with dMMR or MSI-H status.
The approval of dostarlimab was based on data for patients with dMMR or MSI-H status. However, the manufacturer did not select patients by mutation status when compiling comparator data from registries. However, to assess the comparability of both populations, the MMR/MSI status is a relevant criterion. Moreover, this biomarker is also an important prognostic factor.
Lanadelumab – Study populations too different
Lanadelumab is a routine prophylactic treatment to prevent hereditary angioedema attacks.
The approval was based on a placebo-controlled trial. For the benefit assessment, the manufacturer aimed to compare individual patient data (IPD) from the approval study with their own studies for the ACT.
However, these study summaries were not suitable for the benefit assessment. The study populations for both arms were completely different, and statistical methods could not compensate for these big differences.
Further reading related to RWE
- What’s important to know about real-world evidence data collection in Germany?
- IQWiG flags up legal barriers to registry-based RCT
- How will German early benefit assessments change in the future?
Newsletter
If you liked this post and don’t want to miss the next one, sign up for my free weekly newsletter!