>> The background to the application-accompanying data collection now introduced by politicians, which is already called ABDE in G-BA parlance, is, according to Müller, due to the significantly increasing number of accelerated approval procedures. In these procedures, the approval authorities enabled market access based on a purely weighing decision, given that the data available at the time was limited and often immature. This means, among other things, that the BMG sees the resulting tendency in the Federal Committee's resolutions that the category “unquantifiable” is increasing sharply. Müller described this as a “pain point”. He could understand these decisions - which came about due to the limited data available - but they did not correspond to the mandate given by the legislature to the Federal Committee, "which should give an assessment for patients and for price negotiations of new drugs as to the probability of an expected effect occurs or not”.
Based on the fact that six levels of evidence are listed in the Drug Benefit Assessment Ordinance, according to Müller, the BMG expects that the Federal Committee and IQWiG will work together to find a methodology to reflect these benefit categories in decisions in order to assess the effects that patients have of new drugs expect to be able to quantify. He clearly addressed the G-BA directly by saying verbatim: “If the Federal Committee only ever decides on 'not quantifiable' for many products, the indicator and the guideline that the legislature had in mind are missing at this central point Federal Committee promises.”
Now, according to Müller, the possibilities for generating the missing data are relatively weak on the part of the European approval. This is exactly what the ABDE, which has been possible since August 16th of this year, is intended to change. The declared goal: to generate a new, “very good level of data” through a link with the service providers and the real care in order to then make the best possible decisions on this basis. For this reason, according to Müller, the ABDE instrument was introduced “admittedly painfully”. But this also happened because there were consequences in the price negotiations with possible market exits, precisely because the G-BA did not come to a quantification in its decisions. Müller: “We know that this is a turning point, but we hope that both or all sides deal with it responsibly.” According to Müller, that would at least be the idea. Whether this works will now have to be found by the system.
This “system” is formed primarily by the G-BA and secondly by IQWiG. In her lecture, which Dr. Antje Behring had titled “Are conditional approvals ready for early benefit assessment?” and signed it with the subtitle “Dealing with evidence gaps in additional benefit assessment”, she answered quite sybillically: “I would like to answer the question very clearly - yes, sometimes no or sometimes more, sometimes less.” The problem that politicians want to address with the ABDE’s initiative is not a new one, because various drugs have already been evaluated in the early benefit assessment within the scope of the conditional approval.
According to Behring, one of the first active ingredients was crizotinib, for which comparative studies were already available in 2013 in the first conditional approval procedure. At that time, the evaluable data cut would have been sufficient to quantify an additional benefit. When the relevant studies were completed in the second step and full approval was achieved, the additional benefit could have been confirmed.
This is different with the active ingredient ceritinib, for which there was only a single-arm study in 2015, at the time of the first benefit assessment, which, according to Behring, led to the additional benefit being determined to be “not proven”. However, with “full approval” in 2017, a “significant additional benefit” could have been recognized through the comparative study that was available.
A similar picture emerged with the active ingredient osimertinib, which was first given an “unproven” additional benefit in 2016, but then with a “considerable” one in the early benefit assessment in 2017. The problem here was that after the initial assessment, in which only data from single-arm studies were available, the pharmaceutical company that marketed it decided to withdraw the drug from the market. Which would have meant nothing other than that the active ingredient was not available in Germany - apart from individual cases - until the second evaluation was successful.
In her presentation, Behring cited blinatumumab (2016 “not quantifiable”, 2017 “significant additional benefit”) and the active ingredient alectinib, for which a hint of “minor additional benefit” was decided as further examples of benefit assessments after conditional approval. Behring: “Alectinib is an example of the fact that at the time of the conditional marketing authorization there were sufficient studies available to quantify an additional benefit.” Here it was the case that the corresponding phase III studies had not yet been completed at the time of approval were, however, available at the time of the AMNOG assessments, which is why they could then also be taken into account.
According to Behring, the situation is different with the active ingredient olaratumab, which was recognized as having “significant additional benefit” in 2017. The evidence available at the time of approval was actually very convincing, as very convincing survival benefits were shown in a comparative phase II study. Behring: “You would never have thought in your life that these data could not be confirmed.” However, this happened in the following phase III study (“Announce”), which could not confirm the survival benefits originally described, resulting in approval in 2019 was withdrawn.
According to Behring, this shows an uncertainty inherent in the available data. She openly admitted that she did not know what would have happened if the phase III study for olaratumab had not been initiated, but instead data from the care had been used, which might have yielded different findings and possibly led to this In the end, the approval did not have to be withdrawn.
This again points to the problem of conditional approvals with their limited data, which leads Behring to question what kind of new evidence is actually required after approval. Here she formulates the key question: “If you assume that the necessary evidence for conditional approvals is already provided by the approval, why do you need evidence from care-related studies?” Or shouldn’t it be better, with a particular focus on the example of olaratumab , further comparative studies will be required by the regulatory authorities?
Since Germany cannot enforce this on its own in a global context, politicians have reacted and introduced the ABDE for Germany with the GSAV. In Behring's opinion, the measure for data collection accompanying the application, which will soon be activated, “can also take place before the decision on the early benefit assessment, in parallel or even afterwards”. This should not be confused with the concept of time limits, under which pharmaceutical companies have long been required to collect certain data. Behring: “The ABDE is something that can possibly take place independently of it.” In addition, the imposition of the ABDE is “possible, but not mandatory” according to the will of the legislature.
According to Beh-ring, the G-BA will have to analyze very carefully in the future where it actually makes sense to require additional data collection. Behring: “This is certainly not the case with drugs for which we have enough evidence and can already quantify it with the available data.” In her opinion, all active ingredients that do not necessarily research patient-relevant endpoints in the approval studies would fall into a second category have. The G-BA wants to see “major milestones within a disease” in the ABDE to be imposed, because it is not about “small changes” that may no longer be of interest later in the benefit assessment. Her advice: “You would like to know what an active ingredient actually does for the patient; This means that you have to concentrate on good endpoint surveys when collecting data during the application.”
In Behring's view, a dilemma is the timing. The G-BA would like to start collecting data as part of the ABDE as early as possible. The problem is that the evidence gap must first be defined and described as precisely as possible “which data is actually needed in order to be able to better quantify the additional benefit”. Her question: “Does this only have to be done after the benefit assessment or could the approval authorities not agree beforehand to find out what evidence gap there actually is?” Here she sees the pharmaceutical companies as having an obligation to explain what possible gaps there are, which “ “Doesn’t exactly lead to a flower pot in the benefit assessment.”
Here, however, it is also important to organize the cooperation between the G-BA and the approval authorities in a timely manner in order to agree at an early stage on which data the approval authorities on the one hand and which possibly the G-BA additionally requires on the other hand.
Once the ADBE is imposed, it will depend on where the required data actually comes from. The G-BA thought about the “where”, Behring’s answer: “from registers”. But she now knows very well that the time factor will be crucial here; especially if there is no register at all for a specific data question and no concept of what application-accompanying data collection could look like, that – according to Behring – “takes a while”. As an example of this, she mentioned a registry study to be set up to document certain car T cell therapies, which, at the request of the EMA, was established with the registry of the “European Society for Blood and Marrow Transplantation” (editor's note: the backbone of the system since 1974). Research and educational activities of the EBMT with more than half a million patients who have received a hematopoietic stem cell transplant). Talking to the registry operators began at a very early stage, and the registration and HTA authorities also made an enormous effort to involve stakeholders - and yet the registry study was “incredibly tough and difficult to set up”. Her conclusion from this experience: “This means that if you only impose application-related data collection at the time of the benefit assessment, it may be too late.” The rule here is: “the earlier, the better.”
Another question arises about the type of register, “is it in the direction of product registers or application observations in order to be able to answer specific questions”. She immediately rejects the former. Behring: “The short-term, temporary establishment of product registers is not exactly what we would like.” In her opinion, work should be done on how questions from early benefit assessment could be integrated into potentially existing disease-specific registers, without using different data sources in parallel must be operated. The problem is that indication-related registries do not collect all the data that are suitable for certain questions within the benefit assessment.
All of this leads to the question of “How?” According to IQWiG director Wascher, the legislature has made a “very sensible move to stimulate data collection,” but now comes “the comma and then the but.” In his opinion, although this is a very diplomatic formulation, the ABDE is “not yet optimally designed”. Here he primarily thinks about practical feasibility, effectiveness and the necessary development measures.
According to the legal resolution, the central point of data collection during the application, namely generating data to quantify the additional benefit for a benefit assessment or an additional benefit assessment, requires certain procedures and also certain minimum requirements. He is somewhat skeptical as to whether the G-BA has the ability to do what is actually necessary with the existing legal framework, which particularly concerns the exclusion of randomized clinical trials, although he “did not understand the reasoning for this”.
From the two previous lectures by Müller and Behring, he understood, and could also sign, that “the target direction and the very essential data source for the data collection accompanying the application will obviously be registers”. However, he vetoed the term application observation used in the draft law; no one needs it, “neither the term nor the so-called studies” and “certainly not in order to quantify benefits in the early benefit assessment”. There are also wordings in the legal text – “let’s just say that they are difficult to understand” – such as the fact that benefits can be proven through case reports or expert opinions, which is “somehow a completely new finding”.
However, Wascher sees the biggest problem as rooted in a different situation. Given the difficulty of quantification, the priority is not “that we do not know enough about the new drug”, but rather the crucial point is “that we do not have enough data about comparative therapies and the spontaneous progression of the disease”. This means that focusing data collection during use on a new drug “will not help us with the question of the benefit of the effect and the quantification of the benefit”. Hence his demand: “We need the comparison.” For this reason, indication-related data is needed instead of product-related. However, it is now a crucial point that the patient population to be examined is “currently based relatively closely and strictly on the approvals”, which also applies to the application-related data collection. If the G-BA maintains this requirement in the current way, that an approval reference is also necessary in the data collection accompanying the application for registers, this would seem to be “in no way a trivial requirement” given everything it knows about registers.
The same applies to dosages and dosage regimens. If the G-BA maintains existing requirements for these points, they would also have to be reflected in the data collection accompanying the application.
Winderer sees another problem area with therapies, for example with orphans for genetic diseases for which there are no other comparable therapies. Nevertheless, if there are no comparative therapies, you at least need comparison situations, because: “Without comparison, nothing works.” diaperer: “It is a mistake to believe that it is a utopia that one can make a statement about additional benefits without comparison.” That The term additional benefit already says that it implies “compared to something else”. It is therefore crucial to consider how such a comparison can be designed so that it allows the quantification of additional benefits.
However, it also applies that the smaller the differences to be expected, the greater the requirements for the fairness of the comparison. To do this, you always need alignment and the intention-to-treat principle, which you would then also have to implement in registers.
Since none of this would be easy, IQWiG is currently in the process of discussing with various registry operators how registries should be prepared in order to be able to implement comparative studies in registries. This undertaking by his institute was well received by the register operators because they would be interested in their efforts to keep a good register being used for something worthwhile.
But what does a good register mean? According to diaperer, what matters most is what level of demands the G-BA wants to place on their quality. This begins with the question of how complete data should be collected in registers. And this extends to the question of whether the data found in registries even allows quantification, including the associated reference variables, populations, side effects and quality of life. The real world evidence then relies on this data source as a research methodology, described in the course of the MVF congress by Friedhelm Leverkus (Pfizer, Berlin), Prof. Dr. Wolfgang Hoffmann (Institute for Community Medicine, Greifswald), Prof. Dr. Frank Griesinger (University of Oldenburg) and Prof. Dr. Franz Porzsolt (ICE eV, Ulm) first appeared.
from:
MVF editor-in-chief Peter Stegmaier
Citation note:
