Background: HTA methods for evaluating oncology survival benefit were introduced when the predominant drug treatments were cytotoxic regimens for advanced disease, delivering moderate survival benefit in most indications. Immunotherapies now offer the potential for transformative benefit in an increasing number of indications. The availability of adjuvant drug therapies with curative intent in earlier disease stages is also expanding. The increased diversity in the survival benefit delivered by new technologies adds complexity and presents new challenges for HTA evaluation of oncology therapies.

Summary: We separate life-extending therapies into three categories: palliative, curative and mixed. Palliative therapies provide moderate survival gains across the treated population. Curative therapies can produce transformative outcomes and do so for the majority of patients. Mixed therapies can provide transformative benefit, but do so for a minority of patients, with the majority receiving either moderate or no survival benefit.

We identify the different metrics, methods and issues that are relevant for the evaluation of survival benefit for each category. Median survival, hazard ratio and proportional hazards methods are relevant for evaluating palliative therapies. Landmark survival, cure fraction and polynomial models become relevant for mixed therapies. Landmark survival, surrogate endpoint validation and surrogate modelling become key for curative therapies. Issues of equity may arise more readily for mixed therapies. Quality of long-term survival becomes an important consideration for curative therapies.

Recommendations: This conceptual framework can help HTA agencies and developers to focus on the most appropriate metrics and methods and target the most important issues in the assessment of oncology therapies. HTA methodologists should use the framework to guide the development of new tools and methods. The development of objective criteria to define the categories could also assist in the integration of this approach into rules-based value frameworks.