Treatment response in hemato-oncology in the context of the German early benefit assessment of drugs compared to clinical practice

Abstract

High payer deficits resulted in a regulation in 2010, which forced new pharmaceuticals in Germany after market access to undergo an early benefit assessment (EBA) based on patient-relevant outcomes to prove an added benefit compared to existing therapies.^{1, 2} This process had been described in detail elsewhere.³

In hemato-oncology, the heterogeneity of the clinical pictures resulted in a large number of endpoints. Survival was often considered decisive. Even survival was not biased by subjective interpretation, it was only suitable to a limited extent due to various factors, including (i) increasing clinical trial duration as longer survival times were achieved even in the palliative setting; (ii) emerging therapies increasingly focused on early-stage diseases and survival might not be measurable in cancers with good prognosis; (iii) fatal events were rare when cancer prognosis was good, leading to increasing sample size to demonstrate statistical difference; (iv) wider therapeutic options and additional number of treatment lines increased the complexity to isolate the effect of the drug, intensified even more due to bias linked to crossover effects; (v) in settings with long postprogression survival due to effective posttrial therapy or crossover, progression could be used as the primary endpoint.^4-7 This was reflected in the approach of the European Medicines Agency, which had chosen a decisive endpoint other than survival in more than half of the newly approved hemato-oncology drugs since 2009.⁸ However, the value of treatment response endpoints (TREs) in hemato-oncology has not been systematically investigated within the framework of EBA and compared with everyday clinical practice.

All EBAs of hemato-oncological drugs with a resolution of the decision maker (Federal Joint Committee, FJC) were included. Regarding the consideration of the TREs by the FJC, the binomial proportion was determined on an indication-specific basis.⁹ TREs used by the manufacturers to prove an added benefit were described and their significance in clinical practice was presented. The assessment of TREs relevance in clinical practice was derived from the German Onkopedia Guidelines and Oncology Guidelines Program regarding therapy goals, prognosis, and influence on therapy decisions. The acceptance of TREs by the FJC was contrasted with their relevance in clinical practice using Cohen's Kappa and interpreting the results according to Altmann et al.¹⁰

Until end of 2021, there were 72 FJC resolutions for hemato-oncological drugs, of which in 48 cases 64 TREs were used by the manufacturer to demonstrate an added benefit (Table 1). Of these cases, 71% used one, 25% two, and 4% three TREs.

The FJC only accepted a total of nine TREs as patient relevant in eight EBAs. Of the 55 nonaccepted TREs, 29% were primary endpoints in pivotal trials. In total, there were FJC resolutions in 17 hemato-oncological indications. The eight EBAs with accepted patient-relevant TREs were distributed across the following indications: chronic lymphocytic leukemia (CLL), Hodgkin's lymphoma, cutaneous T-cell lymphoma, myelofibrosis, polycythemia vera, and systemic anaplastic large cell lymphoma (Table 1).

The overview showed that the TREs accepted by the FJC either (i) in 6/9 TREs a symptom response associated with the endpoint could be shown (EBAs of Brentuximab vedotin for the treatment of relapsed/refractory Hodgkin's lymphoma, relapsed/refractory systemic anaplastic large cell lymphoma, and CD30+ cutaneous T-cell lymphoma, TRE: complete remission; EBA of Fedratinib for the treatment of myelofibrosis and EBAs of Ruxolitinib for the treatment of polycythemia vera and myelofibrosis, TRE: spleen response) or (ii) in 2/9 TREs a treatment that was burdensome for patients and associated with an increased risk of treatment-related side effects could be prevented by reaching a target value (EBA of Ropeginterferon alfa-2b for the treatment of polycythemia vera, TRE: hematological response; EBA of Ruxolitinib for the treatment of polycythemia vera, TRE: hematocrit control) or (iii) in 1/9 TREs the high healthcare relevance of the effects of the therapy justifies the acceptance (EBA of Idelalisib for the first-line treatment of CLL with 17p-deletion/TP53-mutation in combination with Rituximab, TRE: overall response rate). The most common reason for accepting a TRE was that a symptom response associated with the endpoint could be shown.

Out of 55 not-accepted TREs, the rejection as a patient-relevant endpoint was clearly justified by the FJC in the supporting reasons in 23 cases. It turned out that the reasons for rejection were distributed very differently in the individual indications. In acute lymphatic leukemia (ALL) and acute myeloid leukemia (AML), clear reasons were given for almost all TREs as to why they were not recognized as patient relevant in the EBA. In indications such as CLL and multiple myeloma, the rejection was only clearly justified for a few TREs by the FJC.

A comparison with the TREs accepted as patient relevant by the FJC showed that the endpoint “complete remission” was accepted as patient relevant in three EBAs, whereas in 16 cases, it was not accepted. The cases differed in the operationalization of the endpoint. All nonaccepted TREs were not symptom-related but were recorded exclusively on the basis of laboratory tests or imaging procedures such as PET.

The second main reason for rejection was that the TRE was not validated as a surrogate parameter for other patient-relevant endpoints. The hurdles in surrogate validation as part of the EBA have already been discussed elsewhere.¹¹

The binomial proportion of TREs being accepted depends on indication (Table 1). In 10 of 16 indications, no TRE was accepted by the FJC. Within the indications with accepted TREs the proportion of acceptance ranged between 8.3% in CLL and 100.0% in myelofibrosis. Overall indications, the binomial proportion was 0.14 (95% confidence intervals (CI) 0.07 to 0.25). With the exception of the indications of ALL, CLL, and multiple myeloma, the 95% CIs included 0.50 due to the relatively small numbers of TREs observed and the strong heterogeneity between the indications (chi-square test = 36.417, df = 15, p = 0.002).

Of the 11 cases with agreement, five endpoints were accepted as part of the EBA and, according to the guidelines, were also relevant in clinical practice. Six endpoints were not accepted by the FJC as patient relevant and were not listed in the guidelines (Table 1). Except for two cases, the decisions of the FJC were consistent; that is, a specific TRE was accepted or rejected in one indication regarding all EBAs within this indication. In the indication of systemic anaplastic lymphoma, the FJC accepted the endpoint of complete remission in one case but not in another case. In the EBA of brentuximab vedotin (systemic anaplastic large cell lymphoma; first line; combination with cyclophosphamide, doxorubicin, and prednisone), in which the complete remission was not accepted, there was a significant difference across the entire patient population, but the difference was not statistically significant in patients with B symptoms at baseline.¹²

Of the 18 TREs with disagreement on the value in the EBA compared to the clinical practice, 17 were recognized by the FJC as not being patient-relevant. In contrast, these endpoints were listed in the guidelines as therapeutic goals or were relevant for prognosis or the treatment decision. Only reduction of spleen volume by ≥35% in polycythemia vera was accepted by the FJC, which was not listed as a relevant endpoint in the guideline.

Summing up the TREs in the various indications, 47 of 64 TREs were not accepted in the respective EBAs, which according to the guideline were important in clinical practice. With a Cohen's Kappa = 0.01 (standard error = 0.04, 95% CI −0.07 to 0.08), there was only a poor agreement between accepted TREs in EBA and their value in clinical practice. In addition, the respective odds ratio of TREs acceptance in clinical guidelines compared to EBAs indicated with 1.17 (95% CI 0.13 to 10.79) statistical independence between both.

It became apparent that the EBA in Germany and their preceding market authorization as well as the guidelines often came to different conclusions on the value of endpoints, although they were usually based on identical clinical evidence. This was also discussed by relevant medical societies and also applied at the European level.^{13, 14} TREs had a relevant meaning in clinical practice in many therapeutic areas and especially in hemato-oncology as overall survival oftentimes could not be captured in regular study duration. For example, the median survival time for multiple myeloma had more than tripled over the last 20 years to almost 70 months.¹⁵ Minimal residual disease was an important prognostic factor in the treatment of ALL, CLL, and multiple myeloma. Achieving a complete remission was often the primary therapy goal in the treatment of ALL, AML, or diffuse large B-cell lymphoma, because this moved patients usually on to follow-up care and there was no need for any further burdensome treatment associated with side effects. In CLL, a change in therapy was necessary if no complete or partial remission was achieved after first-line therapy. In Hodgkin's lymphoma and mantle cell lymphoma, allogeneic transplantation, a potentially curative treatment, was an option for relapsed patients if they had achieved (partial) remission.

The question arises whether patient-relevance was defined too narrowly in hemato-oncology in the context of EBA ignoring the value of TREs in clinical practice. It is therefore crucial that the chosen endpoints are meaningful to clinicians, patients, and policymakers that are the end-users of evidence generated by these trials.