TA903 · STA

Darolutamide with androgen deprivation therapy and docetaxel for treating hormone-sensitive metastatic prostate cancer

RecommendedTechnology Appraisal Committee BMay 2023

Darolutamide is only recommended if the company provides it according to the commercial arrangement (simple discount patient access scheme)

Source documents

Intervention

darolutamide (not stated)
androgen receptor inhibitor · androgen receptor inhibitor · oral

Condition

hormone-sensitive metastatic prostate canceroncology · metastatic

Comparators

NameType Established Committee preferred
adt aloneactive drugYes
adt with docetaxelactive drugYesYes
adt with enzalutamideactive drugYesYes
enzalutamide plus adtactive drugYes
adt plus docetaxelactive drugYes

Clinical trials

TrialDesignPhasePivotal
ARASENSRCTPhase 3Yes
ARCHESRCTPhase 3
CHAARTEDRCT3
GETUG-AFU15RCT3
LATITUDERCT3
STAMPEDE-2RCT3
STAMPEDE-3RCT3
STAMPEDE-4RCT3

Economic model

partitioned survival (company)
Time horizon: 34 years
Cycle length: 28 days

ICER

Below £20,000 (darolutamide plus ADT and docetaxel vs best supportive care/standard of care) · moderate uncertainty
Below £20,000 · high uncertainty

Methodological decisions (15)

indirect comparison method

Network meta-analysis included abiraterone (not a NICE-recommended comparator) and heterogeneous PFS definitions across trials

Company: Base-case used ARASENS time to CROD and closest matching definitions from other trials; alternative analysis explored PFS definitions not including death

ERG: Scenario analysis removed abiraterone from network; identified inconsistency particularly from STAMPEDE-4 contributing more to residual deviance

Committee: Considered it appropriate to explore impact of removing abiraterone; acknowledged uncertainty about consistency between direct and indirect evidence

ICER impact: uncertain_direction

indirect comparison method

Network meta-analysis structure and consistency. EAG scenario analysis removed abiraterone (not a NICE-recommended comparator) from the network. EAG also noted inconsistency with STAMPEDE-4 contributing higher residual deviance.

Company: 6-trial network meta-analysis including abiraterone. Fixed-effects for overall survival, random-effects for PFS. Used hazard ratios from overall population in ARCHES despite 17.8% prior docetaxel exposure.

ERG: Preferred scenario removing abiraterone and treating STAMPEDE-4 as subset of STAMPEDE-2/3. Noted inconsistency in network and differences in progression-free survival definitions across trials.

Committee: Studies in the network meta-analysis were appropriate despite uncertainty about consistency between direct and indirect evidence. Network meta-analysis was appropriate for decision making.

ICER impact: uncertain_direction

indirect comparison method

Treatment switching in ARCHES (31%) and LATITUDE (12%) from placebo to active comparators after unblinding. Whether to adjust hazard ratios for this switching.

Company: Did not adjust for treatment switching. Unadjusted hazard ratios better reflect clinical practice because people in control arm would subsequently receive an anti-androgen, and a second anti-androgen is not likely to add benefit.

ERG: Suggested separate adjustment for treatment switching in ARCHES and LATITUDE via scenario analysis, as unadjusted hazard ratios may overestimate relative benefit of darolutamide.

Committee: Unadjusted hazard ratios may better reflect clinical practice because people in placebo arm would subsequently have a first anti-androgen, and a second anti-androgen is unlikely to add clinical benefit. Unadjusted hazard ratios appropriate to use.

ICER impact: decreases

indirect comparison method

Impact of removing abiraterone from the network meta-analysis

Committee: noted uncertainty around this impact on cost-effectiveness

population generalisability

ARASENS trial had mostly patients with metastatic disease at diagnosis and high ECOG PS score of 0, with fewer patients of Black/African American background than NHS practice

Company: Demographics applicable to overall NHS population

ERG: Noted high percentage of metastatic disease at diagnosis was generalisable; noted ECOG PS 0 associated with better prognosis but mitigated by metastatic disease at diagnosis; noted under-representation of Black/African American patients

Committee: Overall ARASENS is generalisable to NHS clinical practice; some geographical areas with more diverse populations may be under-represented

ICER impact: negligible

surrogate endpoint validity

Time to CROD (castration-resistant prostate cancer or death) was used as a proxy for progression-free survival in the economic model, rather than radiological PFS

Company: Time to CROD better reflects clinical practice as imaging done based on clinical factors (e.g. PSA progression) rather than fixed schedule

ERG: EAG agreed that time to developing hormone-relapsed prostate cancer was an appropriate outcome and time to CROD was an appropriate proxy for PFS in the model

Committee: Committee agreed time to CROD was appropriate for decision-making but acknowledged importance of considering heterogeneity of PFS definitions across studies. Committee noted company did not have comparison data between time to CROD and radiological PFS

ICER impact: uncertain_direction

surrogate endpoint validity

Assessment method for progression-free survival in ARCHES: centralised independent review (original estimate) vs investigator assessment (updated estimate). Large difference in hazard ratios (HR 0.39 vs HR 0.63).

Company: Used updated radiological progression-free survival estimates from ARCHES (investigator assessment) at technical engagement because they matched ARASENS follow-up timing.

ERG: Did not use updated estimates in base case due to unexplained notable difference in hazard ratios. Original centralised assessment usually more conservative. Noted difference in assessment method may explain discrepancy.

Committee: Preferred latest progression-free survival estimates from ARCHES and STAMPEDE-2, though acknowledged uncertainty around the reason for large difference between interim and final assessments. Clinical experts noted investigator assessments reflect NHS clinical practice decision-making.

ICER impact: uncertain_direction

survival extrapolation

Choice of parametric distribution for overall survival and time to CROD extrapolation over 30 years. Concerns about clinical plausibility of long-term estimates.

Company: Originally used log-normal for overall survival and generalised gamma for time to CROD.

ERG: Preferred log-logistic for overall survival as more conservative. Clinical advice suggested log-normal overall survival estimates over 30 years were optimistic. Updated company to use log-normal for time to CROD.

Committee: Log-logistic for overall survival aligned with after technical engagement. Concluded estimates up to 10 years consistent with STAMPEDE-3 evidence, but beyond 10 years progression-free survival estimates above clinically expected levels. Unlikely people would be progression-free over 20-30 years. Optimistic extrapolations added to uncertainties.

ICER impact: increases

survival extrapolation

Progression-free survival estimates from ARCHES and STAMPEDE-2 trials

Committee: preferred latest progression-free survival estimates from ARCHES and STAMPEDE-2

ICER impact: uncertain_direction

treatment effect waning

Whether treatment benefit with darolutamide changes over time on a population level, particularly beyond 10 years of extrapolation.

Company: Did not explore treatment-effect waning.

ERG: Previous NICE guidance explored treatment-effect waning in scenarios.

Committee: Clinical experts questioned clinical plausibility of waning given more mature ARASENS trial with longer follow-up and comparator including docetaxel. Cancer Drugs Fund clinical lead noted darolutamide continued until progression, so waning may already be captured. Committee acknowledged model may already capture gradual loss in treatment effect. Agreed scenarios exploring further impacts would be useful. Unknown impact of treatment-effect waning added to uncertainties.

ICER impact: uncertain_direction

treatment effect waning

Potential impact of treatment-effect waning on cost-effectiveness noted as uncertain

Committee: noted uncertainty but did not specify preferred approach; acknowledged this as a cost-effectiveness uncertainty

treatment sequencing

Use of second anti-androgen in ARASENS trial not reflective of NHS clinical practice where anti-androgens typically used only once

Company: Company did not adjust overall survival for subsequent anti-androgen treatment

ERG: EAG agreed second anti-androgen unlikely to have clinical benefit

Committee: Noted overall survival results suggest treatment benefit; no adjustment necessary as second anti-androgen unlikely to provide clinical benefit

ICER impact: negligible

utility source

Source of health-related quality of life utilities. ARASENS did not collect EQ-5D data.

Company: Used utilities from enzalutamide guidance (ARCHES and AFFIRM trials).

ERG: Preferred utilities from enzalutamide guidance. Added 0.02 disutility for docetaxel use for 6 months in line with apalutamide guidance.

Committee: Utility values used by company and EAG were appropriate for decision making.

ICER impact: negligible

utility source

Docetaxel disutility adjustment for proportion of people alive during 6 months

Company: applied adjustment for proportion alive during 6 months

Committee: committee noted the company's modelling approach was appropriate, with negligible impact on ICER

utility value choice

Docetaxel disutility application. Whether to adjust disutility for proportion of people alive during 6-month treatment period.

Company: Used 0.02 disutility for 6 months and adjusted for proportion alive during the period at technical engagement.

ERG: Added 0.02 disutility for 6 months without additional adjustment.

Committee: Company's adjustment had negligible impact but modelling approach was appropriate.

ICER impact: negligible

Evidence gaps

no direct comparisonNo direct comparison between darolutamide plus ADT and docetaxel with enzalutamide plus ADT; indirect comparison via network meta-analysis with potential inconsistencies
short follow upLimited radiographic progression-free survival data; time to CROD (composite outcome including PSA progression) used as proxy but no direct comparison data with radiological PFS
otherHeterogeneous definitions of progression-free survival across trials in the network meta-analysis
no direct comparisonIndirect comparison via network meta-analysis used because darolutamide plus ADT and docetaxel compared against multiple active comparators (enzalutamide plus ADT, ADT plus docetaxel, ADT alone).
surrogate not validatedProgression-free survival definitions differed substantially across trials in network meta-analysis (time to CROD, time to clinical progression, radiological progression-free survival, failure-free survival). Large unexplained difference in ARCHES progression-free survival hazard ratios between centralised and investigator assessment methods.
short follow upConcerns about clinical plausibility of modelled overall and progression-free survival estimates beyond 10 years. Unlikely that people would be progression-free over 20-30 years in clinical practice.
otherClinical plausibility of modelled overall and progression-free survival estimates uncertain
otherReason for large difference in treatment effectiveness between interim and final assessments uncertain
otherPotential impact of removing abiraterone from network meta-analysis unclear

Commercial arrangement

simple discount pas · confidential · critical for recommendation

Special considerations

Innovation acknowledged Equality issues raised

Cross-references

comparator guidance — Docetaxel with ADT listed as treatment option per NICE prostate cancer guideline
TA916comparator guidance — Enzalutamide with ADT for treating hormone-sensitive metastatic prostate cancer
TA923comparator guidance — Apalutamide with ADT for treating hormone-sensitive metastatic prostate cancer (used when docetaxel unsuitable)