Lurbinectedin and tarlatamab: What the G-BA benefit assessments reveal after the JCA

The German Federal Joint Committee (Gemeinsamer Bundesausschuss, G-BA) published its benefit assessments of lurbinectedin (Zepzelca) and tarlatamab (Imdylltra) on 1 October 2026. They highlight differences between the European Joint Clinical Assessment (JCA) and Germany's early benefit assessment under AMNOG. Although randomised comparative evidence is available for both treatments, the assessments differ in their consideration of patient-reported outcomes, interpretation of safety findings and handling of methodological uncertainty.

Both medicines are used in adults with extensive-stage small-cell lung cancer (ES-SCLC), but at different stages of treatment. The lurbinectedin JCA addressed one PICO and a direct comparison, whereas the tarlatamab JCA covered seven PICOs supported by different direct and indirect evidence approaches.

The first German benefit assessment following a JCA, for tovorafenib (Ojemda), had already highlighted differences between European and national evidence assessment. The two subsequent procedures add experience from randomised studies and provide further insight into Germany's evidence requirements.

Lurbinectedin and tarlatamab: Two different evidence settings in the JCA

The JCAs for lurbinectedin and tarlatamab differed substantially in assessment scope and available comparative evidence. Germany's Institute for Quality and Efficiency in Health Care (IQWiG) acted as an assessor in both procedures.

Lurbinectedin in combination with atezolizumab is authorised as maintenance treatment for adults with ES-SCLC whose disease has not progressed after first-line induction treatment with atezolizumab, carboplatin and etoposide. The European assessment covered one population and one PICO. It relied on the randomised controlled IMforte trial, which directly compared the combination with atezolizumab alone.

Tarlatamab is used as monotherapy following disease progression during or after first-line platinum-based chemotherapy. This bispecific T-cell engager targeting delta-like ligand 3 (DLL3) was assessed across four populations and seven PICOs. Alongside direct comparisons from the randomised DeLLphi-304 trial, the submission included network meta-analyses and a matching-adjusted indirect comparison (MAIC).

The procedures therefore offer a comparison between a relatively narrowly defined question supported by direct randomised evidence and a more complex assessment involving multiple populations, comparators and indirect analyses.

Lurbinectedin: How the G-BA assessed the IMforte trial

One PICO and one direct comparison in the lurbinectedin JCA

The lurbinectedin JCA relied on a direct comparison of lurbinectedin plus atezolizumab versus atezolizumab from the randomised IMforte trial. The European assessment identified limitations concerning the trial population, certainty of the findings and interpretation of individual outcomes.

The trial did not cover the full authorised population. Patients with an ECOG performance status of 2 or higher and patients with central nervous system metastases at baseline were excluded.

Further uncertainty arose from the open-label design and low, imbalanced questionnaire completion rates between treatment arms for patient-reported outcomes.

Two data cuts were available. The JCA assessors used the later cut because the data were more mature. The treatment effect on overall survival was smaller at this cut and did not reach statistical significance.

By contrast, progression-free survival (PFS) showed a statistically significant advantage for lurbinectedin plus atezolizumab over atezolizumab alone.

Statistically significant disadvantages were also observed on selected scales of the EORTC QLQ-C30 and QLQ-LC13 questionnaires, for severe adverse events of CTCAE grade ≥ 3, serious adverse events and treatment interruptions due to adverse events.

The European assessment thus reported both a benefit in delaying progression and disadvantages in patient-reported outcomes and safety.

Why the G-BA excluded several IMforte outcomes

The G-BA did not include several outcomes reported in the lurbinectedin JCA in its German benefit assessment. Key reasons concerned patient relevance under German assessment standards, the operationalisation of patient-reported outcomes and methodological limitations in data collection.

Like the European assessors, the G-BA used the later IMforte data cut. Differences between the assessments therefore arose primarily from outcome selection and methodological interpretation rather than from fundamentally different data cuts.

PFS and tumour response were not considered in the German assessment, in line with the G-BA's assessment approach to the submitted measures. A statistically significant PFS finding therefore did not automatically contribute to the national assessment of additional benefit.

The G-BA regarded the EORTC questionnaires as patient-relevant instruments in principle. However, it considered the submitted time-to-confirmed-deterioration operationalisation inappropriate because treatment and observation durations differed substantially between arms.

Questionnaire completion rates also declined rapidly and differed between arms. The lack of follow-up after treatment discontinuation created a risk of informative censoring. The submitted documents did not adequately identify the reasons for censoring.

The G-BA consequently judged the risk of bias to be high for patient-reported outcomes and adverse-event outcomes.

Disadvantages on individual EORTC subscales reported in the JCA were excluded from the national assessment because of these methodological limitations. Treatment interruptions due to adverse events were also excluded, as this outcome is not used in the German benefit assessment.

The overall assessment therefore retained, in particular, disadvantages in severe adverse events of CTCAE grade ≥ 3 and serious adverse events.

Lurbinectedin has orphan-drug status. Under German law, additional benefit is therefore considered established by virtue of the marketing authorisation. The assessment concerns the magnitude of that benefit, which is classified as non-quantifiable on the available evidence.

The subsequent written comments, oral hearing and final decision will determine how the identified safety disadvantages affect the assessment of the magnitude of additional benefit.

Lurbinectedin illustrates that European and German assessments can differ even when there is only one PICO and one direct randomised comparison.

Tarlatamab: Seven JCA PICOs and a broader national assessment

DeLLphi-304 and indirect comparisons in the tarlatamab JCA

The tarlatamab JCA covered four patient populations and seven PICOs involving several comparators. Populations were differentiated particularly by time to disease progression and eligibility for further platinum-based chemotherapy.

Comparators in the European assessment scope included topotecan, cyclophosphamide/doxorubicin/vincristine (CAV) and platinum rechallenge.

Amgen submitted evidence for all seven PICOs, but the comparisons varied considerably:

• For PICOs 1, 4 and 6, a direct comparison of tarlatamab with topotecan was submitted from the randomised DeLLphi-304 trial.
• Network meta-analyses addressed PICOs 3 and 5.
• A MAIC was submitted for PICO 2.
• For PICO 7, only an indirect comparison for a single outcome was available.

Across PICOs 1, 4 and 6, the direct comparisons produced broadly consistent results, with advantages for tarlatamab in overall survival, patient-reported outcomes and safety.

The indirect comparisons, however, had major methodological limitations. The JCA assessors concluded that the necessary similarity assumptions were substantially violated and considered the certainty of the indirect evidence very low.

The JCA report thus distinguished the relatively informative direct randomised evidence from the much more limited indirect analyses.

The German assessment then examined which DeLLphi-304 findings were usable for the overall authorised population and how closely the study reflected clinical practice in Germany.

Tarlatamab in the G-BA assessment: Survival benefit and uncertainty in the evidence

The G-BA benefit assessment found a statistically significant overall survival advantage for tarlatamab versus chemotherapy. It also identified limitations in the applicability of the findings to Germany and the methodological reliability of individual outcomes.

Unlike the European scope, which distinguished four populations, the national assessment considered the broader population covered by the authorised indication.

In DeLLphi-304, tarlatamab demonstrated a statistically significant overall survival benefit versus chemotherapy (hazard ratio [HR] 0.60; 95% confidence interval [CI] 0.47–0.77; p < 0.001).

There were also statistically significant advantages for severe adverse events of CTCAE grade ≥ 3 (HR 0.46; 95% CI 0.37–0.58; p < 0.001) and treatment discontinuation due to adverse events (HR 0.37; 95% CI 0.19–0.72; p = 0.002).

For patient-reported outcomes, the G-BA considered only a subset of the submitted results.

EORTC QLQ-C30 responder analyses showed a statistically significant advantage for tarlatamab in time to first deterioration of at least ten points on the diarrhoea symptom scale (HR 0.67; 95% CI 0.49–0.92; p = 0.01).

Conversely, the appetite-loss symptom scale showed a statistically significant disadvantage (HR 1.27; 95% CI 1.02–1.59; p = 0.04).

Other patient-reported outcomes were excluded because of methodological limitations, particularly the timing of assessments and completeness of the data.

Six-week assessment intervals were considered too long to reliably rule out deterioration events between visits. For several instruments, completion rates had already fallen below 70% by the first or second post-baseline assessment.

The FACT-G GP5 single item raised additional questions about content validity and other psychometric properties. In particular, it remained unclear whether this item could validly capture the full range of treatment side effects.

As with lurbinectedin, PFS and tumour response were not included in the national benefit assessment.

The G-BA also identified concerns about applicability to German care. DeLLphi-304 was conducted internationally, and almost one-third of patients in the control arm received a treatment not authorised in Germany. This limits interpretation of the comparison in the German treatment setting.

In addition, almost one-third of patients in the tarlatamab arm were treated outside the conditions of the product information because treatment could continue after radiological progression.

The open-label design, differences in treatment duration and declining, imbalanced questionnaire completion rates contributed to a high risk of bias for patient-reported outcomes and adverse events.

The G-BA also noted that follow-up for overall survival was ongoing. At the relevant data cut, 56.3% of patients in the tarlatamab arm and 40.0% in the chemotherapy arm were still participating in the study. The effect estimate could change as the data mature.

The evidence picture therefore differs from lurbinectedin. Whereas safety disadvantages remained central to the national assessment of lurbinectedin, tarlatamab showed statistically significant benefits in overall survival and several safety outcomes.

The final determination of the magnitude of additional benefit remains subject to the subsequent G-BA procedure.

JCA versus G-BA: What differs for lurbinectedin and tarlatamab?

The two assessments show that randomised comparative evidence does not guarantee identical conclusions at European and national level. Differences arise from assessment scope, outcome selection and operationalisation, and requirements for certainty of evidence.

The treatment of patient-reported outcomes is a particularly important point of difference.

For both medicines, the JCA reported results that the German assessment did not consider, or considered only in part. Although the details differed, recurring methodological issues included assessment intervals, questionnaire completion rates, observation periods, censoring and instrument validity.

PFS and tumour response illustrate another recurring difference. These outcomes may fall within the European assessment scope without satisfying German requirements for patient relevance and operationalisation in the submitted form.

For tarlatamab, applicability to German clinical practice was an additional concern. The choice of comparators and actual treatment administration influence how the observed effects can be interpreted nationally.

The two procedures therefore expose different challenges at the JCA–AMNOG interface. For lurbinectedin, the key issue is the methodological usability of individual outcomes. For tarlatamab, the reliability and applicability of statistically significant benefits in German care are also central.

Why the JCA reports were not included in the initial German benefit assessments

According to the underlying manuscript, the final JCA reports for lurbinectedin and tarlatamab were not available in time to be considered in the initial German benefit assessments. The national assessments relied on the submitted dossier materials, including references to the European JCA dossier.

A distinction is needed between the manufacturer's JCA dossier and the final JCA report.

The JCA dossier contains the study information, analyses and results submitted for European assessment. Under the applicable conditions, these materials may be reused in the national AMNOG dossier or incorporated through specific cross-references.

The final JCA report, by contrast, documents the findings and methodological observations of the European joint clinical assessment.

The publication date of the European assessment report and the stage reached in the German procedure therefore matter for the practical connection between JCA and AMNOG.

The lurbinectedin and tarlatamab procedures illustrate a timing challenge alongside the differences in evidence requirements. Access to European dossier materials does not necessarily mean that the final European assessment report is already available for the national assessment.

What lurbinectedin and tarlatamab mean for future AMNOG assessments

The two procedures show that European joint clinical assessment can identify important evidence uncertainties, while the usability of the results for German benefit assessment is still evaluated independently.

Both JCA reports address methodological limitations, including open-label trial designs, incomplete patient-reported data and the limitations of indirect comparisons. The German assessments consider similar issues but apply national standards when selecting and operationalising outcomes.

For lurbinectedin, several results presented in the JCA do not contribute to the national overall assessment, leaving safety disadvantages as a central finding.

For tarlatamab, statistically significant benefits in overall survival and severe adverse events remain, although differences from German clinical practice and methodological uncertainties affect their interpretation.

For pharmaceutical manufacturers, this creates a practical evidence-planning consideration: outcomes, assessment intervals, observation periods, comparators and analyses need to be evaluated early against the requirements of subsequent national benefit assessments. See the JCA dossier guide and the AMNOG process guide.

These initial procedures provide practical experience of how JCA and AMNOG interact. They show which evidence can be assessed jointly at European level and where national assessments may still select and interpret results differently.

It is too early to draw a definitive conclusion about the extent to which JCA will reduce the burden of national assessment procedures over time.