The first four published Joint Clinical Assessment (JCA) reports reveal recurring methodological challenges in European evidence assessment. Key issues include PICO scoping and comparators, indirect treatment comparisons, the selection of prognostic factors and effect modifiers, and the handling of non-randomised evidence and post-hoc analyses. For manufacturers, the critical question is therefore whether the evidence for each PICO can be planned, generated and presented in a methodologically robust and transparent way.
The first four JCA reports reveal recurring methodological challenges
Four JCA reports have now been published for tovorafenib, lurbinectedin, tarlatamab and onasemnogene abeparvovec, covering very different evidence settings. These range from randomised direct comparisons to indirect treatment comparisons, single-arm studies, historical controls and non-randomised evidence.
Despite these differences, recurring methodological issues are emerging. A systematic review of the four assessments across key dimensions such as PICO adherence, systematic literature reviews, study design, risk of bias, direct and indirect comparisons, prognostic factors, outcome definitions and subgroup analyses highlights four areas in particular:
PICO and comparator selection determine the comparative evidence required.
Indirect comparisons are scrutinised against their underlying methodological assumptions.
The selection of prognostic factors and effect modifiers affects the robustness of adjusted comparisons.
Pre-specification and transparent analytical decisions influence which results can ultimately be considered in the assessment.
Four JCA reports are not sufficient to draw definitive conclusions about how European assessment practice will evolve over time. They do, however, provide concrete indications of methodological requirements that manufacturers can already incorporate into future JCA planning.
PICO and comparators determine the evidence strategy for a Joint Clinical Assessment
PICO scoping translates a marketing authorisation indication into specific assessment questions covering population, intervention, comparator and outcomes. The greater the differences in relevant populations and comparators across Member States, the more distinct evidence questions may need to be addressed within a JCA.
The first JCA reports illustrate the practical consequences. Eight PICOs were defined in the assessment scope for tovorafenib, of which two were addressed in the submitted JCA dossier. For tarlatamab, all seven PICOs were addressed. For onasemnogene abeparvovec, seven of 14 PICOs were addressed.
The number of PICOs alone, however, only partly captures the resulting workload. Different comparator scenarios can require separate evidence searches, direct or indirect comparisons and additional analyses.
Evidence strategy for a JCA therefore begins well before dossier development. Manufacturers need to anticipate which populations and comparator scenarios may emerge from European scoping and determine whether suitable evidence is available for these potential scenarios.
Indirect comparisons in JCA are scrutinised against their underlying assumptions
Network meta-analyses (NMAs), matching-adjusted indirect comparisons (MAICs) and other indirect treatment comparisons can address evidence gaps when direct randomised evidence is unavailable for individual PICO questions. The first JCA reports show, however, that technically conducting an indirect comparison is not sufficient.
Unanchored MAICs played a role in the assessments of tovorafenib and tarlatamab. Scrutiny included whether relevant prognostic factors and effect modifiers had been identified and appropriately accounted for. For tarlatamab, different methods were used depending on the PICO and comparator, including direct evidence, NMA, unanchored MAIC and an anchored indirect treatment comparison.
The assumptions underlying an indirect comparison are therefore central to its interpretation. For an NMA, this includes the comparability of the included studies. For an unanchored MAIC, it becomes particularly important whether relevant prognostic factors and effect modifiers have been comprehensively identified and accounted for.
More sophisticated statistical methods cannot automatically overcome structural limitations in the underlying evidence. Lack of randomisation, unmeasured confounding or limited overlap between populations remain important sources of uncertainty even after statistical adjustment.
Prognostic factors and effect modifiers are critical to the robustness of MAICs
Identifying and selecting prognostic factors and effect modifiers is a central methodological step in adjusted indirect comparisons. The first JCA reports show that assessors examine not only the statistical adjustment itself but also how the variables included in that adjustment were identified and selected.
This becomes particularly problematic when the literature search used to identify relevant variables is incomplete or when the selection of individual variables is insufficiently justified.
For an unanchored MAIC, this issue is especially important. Without a common randomised comparator, the validity of the comparison depends heavily on whether relevant differences between populations can be adequately accounted for. Unobserved or unmeasured confounders cannot be eliminated through adjustment.
Effective sample size (ESS) also needs to be interpreted in the context of the underlying populations. Reporting a formal ESS alone does not establish whether weighting has left a sufficiently comparable and informative population.
For manufacturers, the methodological justification for a MAIC therefore starts before modelling. Identifying effect modifiers and prognostic factors, providing a clinical and methodological rationale for their inclusion, and documenting the variable-selection process all form part of the evidence strategy.
Pre-specification influences which evidence can be considered in the JCA
The first JCA reports also demonstrate the importance of pre-specified analyses and transparent analytical decisions.
For lurbinectedin, post-hoc selected immune-related adverse events were not considered because the selection was regarded as insufficiently comprehensive. For tarlatamab, certain subgroup analyses based on mixed models for repeated measures and generalised linear mixed models were considered implausible and excluded from further consideration.
These examples illustrate a practical consequence: additional analyses do not automatically strengthen a JCA dossier. What matters is whether their selection, methodology and implementation can be transparently justified.
Pre-specification, transparent statistical analysis plans and robust justification of post-hoc analyses therefore become increasingly important components of evidence planning.
The four JCA reports cover very different evidence settings
The four published JCA reports are particularly informative because they represent different starting points for evidence assessment.
Tovorafenib: PICO fragmentation and limited comparative evidence
For tovorafenib, a single-arm evidence base encountered multiple PICO questions and different comparator requirements. Some of the resulting evidence gaps had to be addressed through indirect comparisons.
The case illustrates how strongly the assessability of evidence depends on the availability of suitable comparative data for the populations and comparator scenarios defined in the assessment scope.
Lurbinectedin: data maturity and pre-specified analyses
For lurbinectedin, randomised direct evidence was available. Methodologically relevant issues included data maturity, different data cut-offs and analytical decisions.
The case demonstrates that randomised evidence does not end the methodological discussion. The timing and planning of analyses and the interpretation of available data remain relevant to the assessment.
Tarlatamab: different evidence methods for different PICOs
Tarlatamab illustrates particularly clearly the analytical complexity that can result from multiple PICO questions. Depending on the population and comparator, the evidence package included direct randomised evidence, NMA, unanchored MAIC and anchored indirect treatment comparisons.
The JCA therefore shows that a single evidence method does not necessarily support an entire assessment. For each PICO, manufacturers need to determine which comparative evidence is available and whether the assumptions required for the respective method are met.
Onasemnogene abeparvovec: evidence assessment in very small populations
Onasemnogene abeparvovec expands the initial JCA experience to an evidence setting characterised by highly specific genetic subpopulations and limited historical comparator data.
In very rare diseases, non-randomised evidence and historical controls may be necessary when suitable direct comparative data are unavailable. At the same time, risk of bias, potential confounding and the assumptions underlying long-term extrapolations become particularly important.
The case therefore highlights a challenge likely to remain relevant for future JCAs in rare diseases: the need to use external comparator data does not reduce the methodological requirements for justifying their use.
JCA dossiers need to be prepared before the final assessment scope
The first four assessments reveal substantial operational as well as methodological requirements. Sections 1 to 6 of the JCA dossiers reviewed range from approximately 208 to 480 pages. At the same time, the number and coverage of PICO questions vary considerably.
The operational challenge arises primarily from the combination of PICO complexity and compressed timelines. Once the assessment scope has been finalised, existing evidence searches and analyses may need to be updated or expanded to cover additional populations and comparator scenarios.
One approach is therefore work-at-risk evidence generation: manufacturers simulate potential PICO scenarios before the final assessment scope is available and prepare evidence searches and analyses for likely scenarios in advance.
This reduces reliance on developing key evidence components only after the final scope has been communicated. It is particularly relevant for systematic literature reviews and indirect comparisons, where methodologically robust execution cannot be compressed indefinitely.
Tovorafenib illustrates the interface between JCA and AMNOG
The German benefit assessment of tovorafenib provides an early practical example of how European and national evidence requirements can differ for the same therapy.
In the German procedure, relevant issues included the patient relevance and operationalisation of endpoints, adherence to the intention-to-treat principle, response rates for patient-reported outcomes and the presentation of safety data.
The handling of indirect comparisons also illustrates the national translation required. MAIC analyses from the European JCA context were mentioned in the German procedure but were not presented accordingly as evidence for the AMNOG assessment.
An evidence package developed for the methodological requirements of a JCA is therefore not automatically suitable for the German benefit assessment. Germany continues to apply the national requirements of the AMNOG process.
For delta dossier preparation, this means manufacturers need to assess early which PICO questions, comparators, endpoints and analyses from the European evidence package also address the requirements of the German benefit assessment and where additional national evidence or analyses may be required.
What the first four JCAs mean for manufacturers
The first four published JCA reports do not yet provide a definitive picture of future European assessment practice. They do, however, reveal recurring requirements that can already inform evidence planning.
For manufacturers, five preparation priorities emerge:
Simulate PICO scenarios early. Potential populations and comparator scenarios should be anticipated before the final assessment scope is available.
Assess comparator evidence early. Gaps in direct comparative evidence may require additional systematic literature reviews and indirect comparisons.
Plan indirect comparisons around their assumptions. Identifying prognostic factors, effect modifiers and potential confounders is part of the methodological preparation.
Pre-specify analyses wherever possible. Post-hoc analyses require particularly robust methodological justification.
Consider European and national requirements in parallel. A JCA evidence package will not necessarily address all requirements of subsequent national HTA procedures.
Clinical development, biostatistics, evidence synthesis and market access therefore need to become more closely connected. The objective is not to generate as many analyses as possible for the JCA. The objective is to prepare a transparent and methodologically robust evidence base for the relevant PICO questions.
Joint Clinical Assessment and national HTA remain different levels of assessment
The Joint Clinical Assessment provides a common European clinical assessment but does not replace national decisions on added benefit, reimbursement or pricing.
The first JCA reports already demonstrate the level of methodological scrutiny applied to the underlying evidence. The first German case also shows that the same evidence must subsequently be interpreted against national requirements.
Manufacturers therefore face two connected tasks: the evidence needs to support the European assessment while being planned in a way that allows relevant national requirements to be considered early.
This is where the delta dossier becomes important. It does not repeat the Joint Clinical Assessment. It connects the European evidence base with the additional requirements of the German benefit assessment.