TA887 · STA

Olaparib for previously treated BRCA mutation-positive hormone-relapsed metastatic prostate cancer

Not recommendedCommittee BAugust 2022

Source documents

Intervention

olaparib (Lynparza)
PARP inhibitor · PARP inhibitor · oral

Conditions

hormone-relapsed metastatic prostate canceroncology · metastatic
brca mutation-positive hormone-relapsed metastatic prostate canceroncology · metastatic

Comparators

NameType Established Committee preferred
cabazitaxelactive drugYesYes
docetaxelactive drugYes
radium-223 dichlorideactive drugYes
best supportive carebest supportive careYes
retreatment with abiraterone or enzalutamideactive drug
abiraterone or enzalutamide retreatmentactive drug
abirateroneactive drug
enzalutamideactive drug

Clinical trials

TrialDesignPhasePivotal
PROfoundRCTPhase 3Yes
CARDRCT3Yes
TAX327RCT3
TROPICRCT

Economic model

markov (company)
Time horizon: lifetime
partitioned survival
Time horizon: lifetime

ICER

Not estimable (olaparib vs cabazitaxel (prior taxane group); olaparib vs best supportive care proxy (no prior taxane group)) · very_high uncertainty

Methodological decisions (32)

comparator selection

Company initially limited appraisal to prior taxane group with cabazitaxel as sole comparator. Committee required consideration of both prior taxane and no prior taxane groups with multiple comparators. Prior taxane group: cabazitaxel, docetaxel retreatment, and radium-223. No prior taxane group: docetaxel and best supportive care.

Company: Limited to prior taxane group; cabazitaxel only comparator

Committee: Separate analysis of prior taxane and no prior taxane groups with multiple comparators (cabazitaxel, docetaxel retreatment, radium-223 for prior taxane; docetaxel and best supportive care for no prior taxane)

ICER impact: uncertain_direction

comparator selection

Whether retreating with abiraterone or enzalutamide is an appropriate comparator. Committee noted this is not offered in NHS practice.

Company: Compared olaparib with abiraterone or enzalutamide retreatment in PROfound

Committee: This comparator does not reflect NHS practice and is not offered in the NHS

ICER impact: uncertain_direction

comparator selection

Use of hazard ratios from BRCA-mutation prior taxane subgroup to model comparative effectiveness with cabazitaxel

Company: Initial approach of comparing BRCA-mutation prior taxane subgroup with whole group

Committee: Use hazard ratios from the BRCA-mutation prior taxane subgroup to model cabazitaxel effectiveness, consistent with use of same subgroup data for other model inputs

ICER impact: uncertain_direction

cost assumption

Method for estimating cost of olaparib using relative dose intensity

Company: Used mean relative dose intensity from PROfound

ERG: Preferred median relative dose intensity; later questioned whether costs should be based on packs prescribed rather than tablets consumed

Committee: Preferred costs calculated for each person based on individual dose and duration; acknowledged company approach acceptable for decision making

ICER impact: negligible

cost assumption

Proportion of patients receiving prophylactic G-CSF with cabazitaxel and duration of treatment

Company: Assumed all people having cabazitaxel had prophylactic G-CSF for 14 days, aligned with CARD and marketing authorisation

ERG: Assumed lower proportion based on clinical expert survey and 7 days duration based on clinical opinion

Committee: Agreed ERG's estimate was appropriate; company followed this approach in revised modelling

ICER impact: decreases

cost assumption

Costing of best supportive care after disease progression

Company: Assumed different costs for BSC depending on whether person had active treatment after progression to avoid double counting

ERG: Assumed same BSC costs regardless of whether active treatment received

Committee: Agreed with ERG approach that same BSC costs incurred for everyone

ICER impact: uncertain_direction

cost assumption

Inclusion of BRCA mutation testing costs in cost-effectiveness estimates

Company: Excluded testing costs in initial base case as test is in NHS Genomic Test Directory; included in scenario analysis

ERG: Included testing costs in base case as NHS does not routinely test; calculated cost to identify 1 BRCA-positive person

Committee: Agreed revised company approach to include testing costs in base case was appropriate

ICER impact: increases

cost assumption

Whether costs of testing for BRCA mutations should be included in cost-effectiveness estimates

Company: Excluded BRCA testing costs from initial base case, included in scenario analysis only, arguing that testing is part of standard NHS practice via the NHS Genomic Test Directory

ERG: Included testing costs in base case, based on clinical advice that routine testing is not standard NHS practice

Committee: Testing costs should be included; testing is not standard NHS care despite inclusion in NHS Genomic Test Directory. Committee noted clinical practice varies and at least one clinical expert does not routinely test unless family history is present.

ICER impact: increases

cost assumption

Cost of best supportive care relative to active treatment receipt

Company: Initially differed cost of best supportive care depending on whether active treatment had been received

Committee: Cost of best supportive care should be the same regardless of whether people had active treatment after progression

ICER impact: uncertain_direction

cost assumption

Use of prophylactic G-CSF (granulocyte-colony stimulating factor) with cabazitaxel

Company: Initially assumed all people on cabazitaxel received prophylactic G-CSF

Committee: Only a proportion of people having cabazitaxel should be assumed to have prophylactic G-CSF, for an average of 7 days

ICER impact: decreases

crossover adjustment

Whether to adjust for treatment switching in PROfound using RPSFTM and recensoring

Company: Used RPSFTM with recensoring to adjust for treatment switching in PROfound; inappropriate to adjust for CARD as trial followed NHS practice

ERG: Agreed RPSFTM was most appropriate; preferred to consider results with and without recensoring as both can bias results

Committee: Company's method for adjusting for treatment switching in PROfound is appropriate including recensoring. Inappropriate to adjust for CARD but uncertainty remains in effect estimate.

ICER impact: uncertain_direction

equivalence assumption

Whether mitoxantrone plus prednisone from TAX327 and TROPIC is equivalent to abiraterone/enzalutamide control arms in PROfound and CARD

Company: Used mitoxantrone as anchor for indirect comparison assuming equivalence to PROfound control

Committee: Mitoxantrone similar to control arms but TROPIC population had not received abiraterone/enzalutamide making it incomparable

ICER impact: increases

indirect comparison method

Company used indirect treatment comparison between PROfound (olaparib vs abiraterone/enzalutamide) and CARD (cabazitaxel vs abiraterone/enzalutamide) to compare olaparib with cabazitaxel. Multiple differences between trials identified.

Company: Used RPSFTM for indirect comparison; BRCA mutation status and prior cabazitaxel do not affect comparability; TROPIC should be excluded

ERG: Highlighted differences in BRCA mutation status, prior cabazitaxel exposure, blinding status between PROfound and CARD. Noted some studies suggest BRCA status modifies treatment effect.

Committee: Differences between PROfound and CARD create uncertainty in indirect comparison; TROPIC population not comparable and unlikely to reduce uncertainty

ICER impact: increases

indirect comparison method

Validity and composition of network meta-analysis for treatment comparisons

Company: Conducted indirect treatment comparison for prior taxane and no prior taxane subgroups

ERG: Participated in network meta-analysis; noted concerns about generalisability of no prior taxane subgroup

Committee: Significant uncertainty in network meta-analyses, particularly for no prior taxane group; this should reduce the maximum acceptable ICER threshold

ICER impact: increases

model structure

Whether to use hazard ratios from whole licensed population or BRCA prior-taxane subgroup when modelling cabazitaxel arm

Company: Initially used hazard ratios from licensed population (larger sample size) rather than prior taxane subgroup

Committee: Should use hazard ratios from BRCA prior-taxane subgroup to match population and other model inputs; committee preferred this revised approach

ICER impact: uncertain_direction

population generalisability

PROfound trial included some people with both abiraterone and enzalutamide prior treatment, which does not reflect NHS practice. Baseline characteristics otherwise generalisable.

Committee: Results should be interpreted with caution; some prior treatment patterns not reflective of NHS practice

ICER impact: uncertain_direction

population generalisability

Whether baseline characteristics in PROfound BRCA-mutation prior taxane subgroup are generalisable to NHS patients. Committee noted some prior treatment regimens (both abiraterone and enzalutamide) did not reflect NHS practice.

Company: Baseline characteristics are generalisable to NHS

Committee: Baseline characteristics generalisable except for some patients having had both enzalutamide and abiraterone before trial

ICER impact: uncertain_direction

population generalisability

Whether post-progression treatments in PROfound and CARD reflect NHS practice. PROfound and CARD included abiraterone/enzalutamide after progression (no clinical benefit in NHS); NHS would use radium-223 dichloride instead.

Company: Excluded abiraterone and enzalutamide from post-progression treatments. Did scenario analyses varying hazard ratio by 5% and 10%.

Committee: Differences in post-progression treatments affect validity of indirect comparison and generalisability to NHS practice

ICER impact: uncertain_direction

population generalisability

Representativeness of PROfound trial population to NHS practice, particularly for no prior taxane subgroup

Company: Used PROfound data for both prior taxane and no prior taxane groups

ERG: Clinical advice suggested PROfound participants without prior taxane were unlikely to represent people who cannot or should not have docetaxel in NHS practice

Committee: Significant concern that no prior taxane group in PROfound does not reflect NHS patients who cannot or should not have docetaxel; no modelling specifically for this clinical scenario

ICER impact: increases

surrogate endpoint validity

PROfound primary endpoint was radiographic progression-free survival; overall survival secondary. Company compared olaparib with retreatment with abiraterone or enzalutamide, which has no clinical benefit and is not offered in NHS practice.

Committee: Comparison with abiraterone/enzalutamide retreatment not reflective of NHS practice; indirect comparison needed with relevant comparators

ICER impact: uncertain_direction

surrogate endpoint validity

In TAX327, company assumed docetaxel's relative treatment effect on progression-free survival equals that on overall survival, as progression-free survival was not reported

Company: Assumed same relative effect size for progression-free and overall survival in docetaxel

Committee: This assumption increased uncertainty in analyses

ICER impact: increases

survival extrapolation

Choice of parametric curve to extrapolate overall survival in prior taxane group. Committee noted none of the curves fitted observed hazard rates well.

Company: Initially chose log-logistic, then exponential citing best fit, then changed to Weibull after committee discussion

ERG: Chose Rayleigh distribution based on best statistical and visual fit; noted Weibull was second preferred

Committee: Both Weibull and Rayleigh appear reasonable but possibly pessimistic; little difference between them; both equally plausible and equally uncertain

ICER impact: uncertain_direction

survival extrapolation

Choice of parametric curve to extrapolate overall survival in no prior taxane group for comparisons with docetaxel and best supportive care

Company: Used log-logistic distribution for both arms

ERG: Preferred Rayleigh distribution

Committee: Both log-logistic and Rayleigh appear plausible; similar extrapolations for docetaxel and best supportive care arms but some differences in olaparib arms

ICER impact: uncertain_direction

survival extrapolation

Choice of parametric curve for extrapolating overall survival in prior taxane group

Company: Initially chose log-logistic, then changed to exponential for best fit; after committee feedback, chose Weibull distribution

ERG: Chose Rayleigh distribution based on best statistical and visual fit

Committee: Agreed both Weibull and Rayleigh extrapolations were equally plausible and equally pessimistic; took both into account

ICER impact: uncertain_direction

survival extrapolation

Choice of parametric curve for extrapolating overall survival in no prior taxane group

Company: Selected log-logistic distribution for both arms; stated it had best visual fit and statistical match to Kaplan-Meier data

ERG: Preferred Rayleigh distribution; considered log-logistic may overestimate long-term olaparib survival

Committee: Concluded both log-logistic and Rayleigh distributions showed good visual and statistical fit with plausible long-term survival predictions

ICER impact: uncertain_direction

survival extrapolation

Choice of parametric curves for extrapolating overall survival beyond trial follow-up

Company: Used Weibull curve in updated base case for prior taxane group; log-logistic for no prior taxane group

ERG: Preferred Rayleigh curve for both groups

Committee: Both Weibull and Rayleigh curves were considered plausible; both predicted at least 3-month survival benefit. However, given uncertainty in network meta-analyses, specific curve choice was not determinative.

ICER impact: uncertain_direction

treatment effect duration

Data source for modelling olaparib treatment duration and costs

Company: Used time until disease progression from PROfound to model treatment duration, as consistent with cabazitaxel trial data

ERG: Preferred to use time to stopping treatment data from PROfound, as curve was above progression-free survival curve and aligned with relative dose intensity

Committee: Agreed that time to stopping treatment better estimates treatment duration and costs of olaparib than progression-free survival

ICER impact: increases

treatment effect duration

Duration of olaparib treatment and timing of dose adjustments

Company: Used time to stopping treatment data from PROfound to model treatment duration

Committee: Time to stopping treatment data is appropriate for modelling duration

ICER impact: negligible

treatment sequencing

Modelling of post-progression treatments after olaparib and cabazitaxel

Company: Model allowed only 1 active treatment after progression; assumed different treatments depending on first treatment; included abiraterone/enzalutamide retreatment; provided scenario analyses excluding cabazitaxel retreatment and all post-progression costs

ERG: Noted company approach did not reflect NHS practice; noted PROfound data showed more than 1 active treatment on average; used trial proportions but acknowledged limitations

Committee: Acknowledged both approaches did not reflect NHS practice but acceptable for decision making as minimal effect on cost-effectiveness

ICER impact: negligible

treatment sequencing

Modelling of post-progression treatments available after olaparib or cabazitaxel failure

Company: Initially included retreatment with abiraterone or enzalutamide after progression

ERG: Challenged the assumption of retreatment with hormonal agents

Committee: Post-progression treatments should not include retreatment with abiraterone or enzalutamide

ICER impact: increases

utility source

Source of utility values for health states

Company: Used utility values from PROfound; mapped EQ-5D-5L to EQ-5D-3L; included quality of life decrement for IV administration of cabazitaxel

Committee: Concluded company's utility values were appropriate

ICER impact: uncertain_direction

utility source

Source and appropriateness of health-state utility values

Company: Used utility values from PROfound trial, mapped EQ-5D-5L to EQ-5D-3L values; modelled worse quality of life with cabazitaxel than olaparib with additional decrement for intravenous administration

Committee: Utility values from PROfound are appropriate

ICER impact: negligible

Evidence gaps

no direct comparisonNo direct trial evidence comparing olaparib with cabazitaxel, docetaxel, or radium-223 dichloride in prior taxane group. Indirect comparison with cabazitaxel uncertain.
no direct comparisonNo direct evidence comparing olaparib with docetaxel or best supportive care in no prior taxane group. Only exploratory indirect comparison available.
short follow upOverall survival data immature in no prior taxane subgroup; small sample size in this subgroup.
no direct comparisonNo direct clinical trial evidence comparing olaparib and cabazitaxel; indirect treatment comparison required
immature overall survivalOverall survival data immature in no prior taxane subgroup of PROfound with small sample size
short follow upAt latest data cut-off (March 2020), PROfound was still collecting data; only 41 events (57%) with olaparib and 27 events (77%) with control in prior taxane group
no uk dataDifferences in post-progression treatments between trial populations and NHS practice
immature overall survivalLimited overall survival events in PROfound at time of analysis; prior taxane group had 41 events (57%) with olaparib and 27 events (77%) with comparator; substantial patient loss after 24 months led to uncertainty in extrapolation
short follow upTrial was still collecting data as planned at March 2020 data cut-off
no direct comparisonNo direct comparison between olaparib and docetaxel retreatment or radium-223 dichloride for prior taxane group; committee could not perform incremental analyses with these comparators
single arm evidence onlyNo randomised controlled trial evidence specifically for people who cannot or should not have a taxane; exploratory analyses used proxy comparators

Commercial arrangement

simple discount pas · confidential

Special considerations

End of life criteria met Equality issues raised

Cross-references

comparator guidance — Abiraterone for treating newly diagnosed high-risk hormone-sensitive metastatic prostate cancer; committee noted similar equality considerations regarding patients unable to receive taxanes
same condition — abiraterone for treating newly diagnosed high-risk hormone-sensitive metastatic prostate cancer
precedent — NICE technology appraisal guidance on cabazitaxel for hormone relapsed metastatic prostate cancer treated with docetaxel; committee considered TROPIC trial in that appraisal
utility reuse — NICE's technology appraisal guidance on cabazitaxel for hormone relapsed metastatic prostate cancer treated with docetaxel; source of utility decrements for adverse events