Surrogates and endpoints in diabetes mellitus

When discussing surrogates and endpoints, one quickly comes to the question of which treatment results and characteristics are actually relevant for the patient. The appropriate expression “patient-relevant endpoint” is anchored in the legal basis for early benefit assessment according to Section 35a SGB V.1 While in other places there is explicit talk of a “therapy-relevant higher benefit” or “therapy-relevant side effects” (cf. e.g. § 35 Para. 1b SGB V), is clearly aimed at the patient. In the associated dossier specifications for Section 35a SGB V, module 3.3.6 mentions patient preferences as a data source to be taken into account.2 Thus, the importance of patient-relevant endpoints and patient preferences has increased with the AMNOG (Law on the Reorganization of the Medicines Market), but what role they play in concrete terms plays is still to be clarified. Nevertheless, the IQWiG (Institute for Quality and Efficiency) commissioned pre-AMNOG for two pilot studies on preference measurement (Conjoint Analysis, Analytical Hierarchy Process), one of which has already been published. IQWiG also focuses on patient preferences in its current method paper (see IQWIG (2013); IQWIG (2010b); IQWIG (2011)).

> While it is hardly possible to objectively determine what should be considered a clinical endpoint and what should be considered a surrogate endpoint, the determination criteria should at least be largely transparent. In addition, there is usually a lack of data to bring more transparency into the relationship (correlation, causality) between the end point and the (supposed) surrogate. This is where positions, value judgments, utility and preferences come into play. Since the health of the patient is of central importance, patient benefits and patient preferences play a central role. This is also reflected in evidence-based medicine (EbM), according to which, in addition to clinical experience, which is the focus, the “clinical condition” and the “patient situation”, “scientific evidence” and “patient preference” are taken into account are (cf. Raspe, H (2007), p.16).
Against this background, numerous questions arise that can only be briefly addressed here: Who actually decides what is relevant to patients? Should patient-relevant endpoints be set by third parties for the patient because they are supposedly patient-relevant? Are all or at least the essential endpoints relevant from the patient's perspective generally recorded in the individual indications? It is undisputed that medical representatives play an important role in determining patient-relevant endpoints. It is therefore to be expected that the majority of patient-relevant endpoints, perhaps even all, have already been determined by medicine. The problem is, we don't know if that's the case. To use the often used comparison with cars: You wouldn't expect drivers to tell the specialist workshop which repairs to carry out, but rather leave this primarily to the specialist workshop. This is due to the information gap that typically exists between the two players, as the majority of car users are only likely to have a layman's knowledge of the technical details of the car (note: However, there is probably a minority of drivers who can define quite precisely what is involved). must be taken into account during the repair) (see, for example, FRITSCH, M et al. (2003)). Despite lay medical knowledge, patients may still have preferences for one or another form of treatment or services or for certain service features. It is not entirely without reason that patients are regularly or irregularly asked by their treating doctors about their preference for one of the treatment alternatives (e.g. in oncology). What drives the patient's decision in favor of one of the treatment alternatives is not always transparent. In principle, however, it can be expected that the preferences of doctors and patients differ sometimes more or less depending on the indication (see, for example, Porzsolt, F. et al. (2010), Danner, M. et al. (2011 )). The differences can be reflected in different weightings of the same endpoints as well as in different endpoints. Furthermore, it can be assumed that tracking the endpoints considered relevant by patients has a positive effect on compliance or adherence. It is important to discuss what consequences this will have for treatment, as well as for HTA evaluations and medicinal products. Mind you, when we talk about patient preferences, what we mean below are the average preferences of a patient group, from which the individual preferences usually deviate to a greater or lesser extent (note: the car experts among the drivers should therefore not be of any importance).
Possible fears that greater consideration of patient-relevant endpoints in the narrower sense and patient preferences will lead to budget-relevant increases in expenditure are likely to be unfounded given the current legal basis. There is the relevant § 12 SGB V, according to which the services must be economical, sufficient, necessary and appropriate.3 Furthermore, lifestyle medicines, where the focus is on increasing the general quality of life, are completely prohibited according to § 34 Paragraph 1 SGB V excluded from coverage by the GKV. Consequently, preferences in the aforementioned sense should only be understood as health-related preferences in analogy to health-related quality of life (HRQoL, Health-Related Quality of Life) (cf. SCHÖFFSKI, O (2007), p.323ff.; EUNETHTA (2012)). Therefore, paternalistic decisions for patients and assumptions about their preferences should fundamentally be replaced by knowledge and concrete data.
The relationship between glycemic control and HbA1c - state of the debate
Although HbA1c is recorded as standard in diabetes studies, blood sugar levels are controversially discussed as an endpoint. It therefore lends itself to a (not complete) overall view as part of this lecture.
In recent years, the results of several larger intervention studies have been published, which have led to a rethinking of treatment goals. The ACCORD, ADVANCE and VADT studies should be mentioned here in particular. Among other things, the effects of a glucose change setting close to normal on micro- and macrovascular events were examined, in which a value of < 6.0% (ACCORD) or < 6.5% (ADVANCE) was monitored. In the ACCORD study, the therapy arm of intensive blood sugar control for type 2 diabetics was stopped prematurely after 3.5 years after increased mortality was shown (cf. ACCORD STUDY GROUP (2008)). The causes of the increased mortality have not yet been clarified. In contrast, the Advance study comes to the conclusion that intensified blood sugar control significantly reduces the incidence of the combined macro- and microvascular endpoint by 10%. While the major microvascular primary endpoint nephropathy is particularly significant, the outcome of major macrovascular events (non-fatal myocardial infarction; non-fatal cerebral insult; death from cardiovascular causes) remains unclear (cf. Advance Collabroative Group (2008)). The current state of research could be summarized as follows: not every patient benefits from aggressive glucose management. Not surprisingly, a glucose change setting close to normal also carries the risk of hypoglycemia (cf. Accord Study Group (2008), Advance Collabroative Group (2008)). Therefore, instead of adhering as strictly as possible to a general HbA1c target value that applies to all patients, it is better to pursue an individual patient-centered approach (cf. Einecke, D (2012); Inzucchi, S. et al. (2012 ); Mehnert, H (2012)). High priority should be given to avoiding side effects (hypoglycemia, significant weight gain) and, if necessary, it is better to leave the blood sugar level at around 7.0% (see DDG (2008)). Consequently, the question is less about whether HbA1c is suitable as a primary or secondary endpoint and more about the rigidity of blood sugar control. In summary, it can therefore be concluded that HbA1C is a surrogate endpoint for glycemic control - however, it is a very well-studied, reliable parameter that is recognized by both international regulatory authorities and specialist societies and is included in the relevant guidelines (see Table 1).
Derivations for evaluation procedures
For example, if you look at the IQWiG assessments since 2006 in the area of diabetes, it is noticeable that different endpoints were used in the final reports; this applies to both the endpoint definition (singular, combined) and the number of endpoints used (cf . Table 2). The endpoints shown in Table 2 are already a summarized version in which, for example, “quality of life”, “health-related quality of life”, “health-related quality of life and therapy satisfaction” and other specifications were summarized under the general term “health-related quality of life”. Understandably, an assessment of type 1 DM in children and adolescents must also take into account physical and psychosocial developmental disorders, which are not given greater importance in the other reports (apart from report A05-01, which is due to the earlier onset of type 1 DM ). However, one would expect a more consistent picture of the endpoints in the other reports.
If you look at the definition of the endpoints hypoglycemia and hyperglycemia in connection with HbA1c, it is noticeable that this varies over time in the benefit assessments (see Table 2). Some of the reports only focus on “severe hypoglycaemia”, others list “hypoglycaemia (particularly severe)” and others list “hypoglycaemia (severe and non-severe)” with equal wording. HbA1c is partly included in the description of the endpoint, but is mostly only listed as a supplement to the endpoints.
In addition, IQWiG has already typified the endpoints in the target variable definition in the two reports A05-07 and A05-10; such typing has not occurred in the other reports.
In report A05-10, most of the endpoints were referred to as “patient-relevant endpoints”, only the endpoints “precursors to blindness: change in the fundus of the eye or visual acuity” and “precursors to end-stage renal disease: change in the glomerular filtration rate (GFR) or serum creatinine” were typed as surrogate endpoints. Consequently, the same typification of the latter two endpoints can also be found in report A05-07. Report A05-07 continues to differentiate between primary endpoints and secondary endpoints. In addition, A05-07 differentiates between “therapy-associated factors” and “consequential complications of diabetes mellitus” (not listed in the table).
Without wanting to give possible reasons for these variations, which would be purely speculative, this shows that determining the end points and defining them is not easy. The inconsistency is possibly a tribute to scientific research by leaving scope for new findings and their influence in evaluations, which is fundamentally positive. For the pharmaceutical industry, this also poses the risk that the endpoints used in the studies do not correspond to the endpoints on which the evaluation procedures are based. In addition, due to a failed validation of surrogates in the meantime, which are used as endpoints in ongoing studies, a later assessment may be correspondingly worse. However, the area of diabetes is very well researched, so there should be broad congruence for the majority of study endpoints and outcomes in the assessment process. In contrast, less researched indications are likely to carry a much greater risk of “endpoint mismatch”. A requirement that could be derived from this is a generally more pragmatic approach to endpoint definition and evaluation.
Patient-relevant endpoints – preference studies by Lilly Deutschland GmbH on diabetes
Lilly Deutschland GmbH has recently carried out two preference studies in the area of diabetes mellitus. In April 2008, a computer-aided conjoint analysis was carried out on 827 adult patients, predominantly with diabetes mellitus type 2 (9% DM type 1) and 60 doctors (30 general practitioners, 30 diabetes specialists) (cf. PORZSOLT, F et al. (2010)). . Using a pre-test, the factors (attributes) of the diabetes treatment were first determined ((main) effectiveness, weight change, application and product type). Here, a largely uniform picture emerges across all patients with regard to the preferences for HbA1c reduction, while the treating physicians' preferences for this were consistently more pronounced. In contrast, the preference for weight loss was approximately the same among obese patients and physicians, while this was less relevant for patients with normal weight or less overweight. Furthermore, the patients rated the absence of side effects as more important than the doctors surveyed.
The second study, a CBCA (choice-based conjoint analysis) on type 1 DM in children and adolescents, was completed at the end of 2012 and is currently in the publication process and can only be roughly outlined here. 152 children and young people between the ages of 8 and 18 were interviewed, in 90% of the cases in the presence of their parents; for some of the young people it also took place without a legal guardian. On the basis of a qualitative survey and with the help of a factor analysis, four attributes (3 times at 3 levels each, 1 time at 2 levels each) were identified which were used within the framework of the CBCA.
Conclusions
HbA1c is a surrogate endpoint of glycemic control, but a globally recognized and well-studied endpoint that can be used, for example, in type 1 diabetics to detect microvascular complications (cf. GLAESKE, G (2012)).
With regard to validation, both the methodology used and the question of who is responsible for validation must be discussed. In principle, there should be a general interest in validating surrogates and endpoints. The data pools of the health insurance companies and the Federal Insurance Office (Section 303e SGB V) could potentially provide important information.
In the author's opinion, an overall consideration of the most important endpoints that are considered important by medical professionals and found to be relevant by patients is fundamentally necessary. To date, the voice of patients has largely been ignored, and for most indications it is not known whether all patient-relevant endpoints have actually been taken into account. How the perspectives could be brought together needs to be discussed together. For the time being, a simple check of existing evidence could be sufficient as part of health technology assessments. <<

Footnotes: see PDF