TA866 · STA

Regorafenib for previously treated metastatic colorectal cancer

Recommended with restrictionsTechnology appraisal evaluation committee BDecember 2022

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

Source documents

Intervention

regorafenib (Stivarga)
tyrosine kinase inhibitor · tyrosine kinase inhibitor · oral

Condition

metastatic colorectal canceroncology · metastatic

Comparators

NameType Established Committee preferred
trifluridine-tipiracilactive drugYes
best supportive carebest supportive careYes
trifluridine–tipiracilactive drug

Clinical trials

TrialDesignPhasePivotal
CORRECTRCT3Yes
CONCURRCT3Yes

Economic model

partitioned survival (company)
Time horizon: 10 years
Cycle length: 1 week
partitioned survival
Time horizon: 10 years
Cycle length: 1 week

ICER

Below £20,000 (regorafenib vs trifluridine-tipiracil) · low uncertainty

Methodological decisions (16)

cost assumption

Relative dose intensity (RDI) modelling approach for regorafenib vs trifluridine-tipiracil

Company: RDI for regorafenib based on mean dose used in CORRECT and CONCUR; for trifluridine-tipiracil, cycle delay and dose reduction modelled separately

ERG: Mean dose from CORRECT and CONCUR already includes both dose delay and dose reduction; real-world evidence (Nakashima 2020) shows similar dose reduction for both treatments (54% vs 48%); EAG preferred equal RDI for both treatments

Committee: Both dose delay and dose reduction should be used for RDI estimation; preferred EAG's approach of applying equal RDI to trifluridine-tipiracil and regorafenib

ICER impact: decreases

cost assumption

Inclusion of grade 1 and 2 adverse events in economic model

Company: Only grade 3 and 4 adverse events (seen in over 2%) captured in base case; scenario analysis with grade 1 and 2 events had negligible impact (fixed cost £5 per event, disutility 0.01)

ERG: Not explicitly stated

Committee: Grade 1 and 2 adverse events should be included in the economic model, despite company scenario showing negligible impact

ICER impact: uncertain_direction

cost assumption

Equal relative dose intensity (RDI) for regorafenib and trifluridine–tipiracil

Company: Company base case did not reflect this preference

Committee: equal RDI for regorafenib and trifluridine–tipiracil

ICER impact: negligible

cost assumption

Inclusion of grade 1 and 2 adverse events in the model

Company: Company base case did not reflect this preference

Committee: include grade 1 and 2 adverse events

ICER impact: negligible

indirect comparison method

Indirect treatment comparison for regorafenib vs trifluridine-tipiracil using network meta-analysis and MAIC, but substantial heterogeneity between trial populations (CORRECT/CONCUR vs RECOURSE/TERRA/Yoshino 2012)

Company: Fixed effect NMA and MAIC show similar efficacy (HR 0.99); potential efficacy modifiers weighted by baseline characteristics

ERG: Concerns about differences in trial populations, particularly progression-free survival differences in Asia-only studies and heterogeneity in prior anti-VEGF treatment exposure and number of previous lines

Committee: Indirect treatment comparison associated with uncertainty due to heterogeneity; regorafenib likely to provide similar benefits but with uncertainty acknowledged

ICER impact: uncertain_direction

indirect comparison method

Use of indirect treatment comparison vs observational evidence to compare regorafenib and trifluridine-tipiracil

Company: Noted high risk of bias in observational studies

ERG: Reported results from observational studies but noted issues with baseline characteristics and potential immortal time bias

Committee: Preferred clinical-effectiveness estimates using indirect treatment comparison over observational studies; noted observational evidence compounds risk of bias due to differences in trial baseline characteristics

ICER impact: uncertain_direction

other

Whether 1.7 severity weighting should apply for trifluridine-tipiracil and best supportive care comparisons

Company: Provided evidence that previously treated mCRC is severe condition; calculated proportional QALY shortfall above 0.95 for both comparators, meeting criteria for 1.7 weighting

ERG: Not explicitly stated as different from company

Committee: 1.7 weighting should NOT be applied for trifluridine-tipiracil comparison due to uncertainty around data used to estimate QALYs (would have preferred real-world data); 1.7 weighting SHOULD be applied for best supportive care comparison as people receiving it would likely be less well with worse prognosis, giving greater face validity to estimates

ICER impact: decreases

population generalisability

Differences in baseline characteristics between CORRECT (global population, more prior biological therapy, more KRAS mutation, more ECOG 0) and CONCUR (Asia-only, longer metastatic disease duration) trials raised concerns about effect modification

Company: Pooled results from both trials are appropriate and generalisable to NHS clinical practice

ERG: Significant differences in trial populations create substantial uncertainty; concerns about anti-VEGF treatment history and baseline characteristics

Committee: Acknowledged differences in baseline characteristics increase uncertainty but concluded pooled results are likely generalisable to NHS practice in absence of further data

ICER impact: uncertain_direction

survival extrapolation

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

Company: Fully parametric models fit to pooled CORRECT and CONCUR trial data; Kaplan-Meier followed by parametric extrapolation for time-on-treatment and PFS

ERG: Preferred fully parametric models to avoid stepped nature of Kaplan-Meier curves causing overfitting; aligns with NICE DSU TSD14

Committee: Generalised gamma is the best visual fit to long-term overall survival data (up to 5 years) provided in clarification response; should be used for cost-effectiveness estimates instead of fitting only to trial data

ICER impact: increases

survival extrapolation

Choice of parametric survival curve for overall survival extrapolation

Company: Fitted parametric models to pooled CORRECT and CONCUR data; used Kaplan-Meier then parametric models for time-on-treatment and PFS

ERG: Preferred fully parametric models for base case, noting that stepped KM curves could cause overfitting; aligns with NICE DSU TA14

Committee: Generalised gamma was the best visual fit to company's long-term overall survival data (up to 5 years) for regorafenib and best supportive care; this should be used for cost-effectiveness estimates

ICER impact: uncertain_direction

survival extrapolation

Committee preferred fully parametric survival models for overall survival estimates using generalised gamma for both regorafenib and best supportive care

Company: Company base case did not reflect this preference

Committee: generalised gamma parametric model for OS

ICER impact: impacted the cost-effectiveness estimates

survival extrapolation

Committee preferred fully parametric survival models for progression-free survival estimates using log-logistic for both regorafenib and best supportive care

Company: Company base case did not reflect this preference

Committee: log-logistic parametric model for PFS

ICER impact: impacted the cost-effectiveness estimates

survival extrapolation

Committee preferred fully parametric model for regorafenib time-on-treatment estimates using log-logistic

Company: Company base case did not reflect this preference

Committee: log-logistic parametric model for time-on-treatment

ICER impact: impacted the cost-effectiveness estimates

treatment sequencing

Post-progression treatment costs and sequencing of regorafenib and trifluridine-tipiracil

Company: Post-progression treatment costs not included in base case; noted that fewer than 10% would have post-progression treatment; scenario analysis applied single cost of £1,633.18

ERG: Not explicitly stated

Committee: Subsequent treatments should be included in cost-effectiveness modelling; around 30% of people would be offered post-progression treatment; no difference in efficacy expected if trifluridine-tipiracil used at fourth line instead of best supportive care (due to different mechanisms of action)

ICER impact: increases

treatment sequencing

Inclusion of subsequent treatments in the cost-effectiveness model

Company: Company base case did not reflect this preference

Committee: include subsequent treatments

ICER impact: negligible

utility source

Source and applicability of utility values for trifluridine-tipiracil comparison

Company: Utility values from pooled EQ-5D-3L results in CORRECT and CONCUR; assumed pre-progression (0.72) and post-progression (0.59) utility values equal for trifluridine-tipiracil and regorafenib because trifluridine-tipiracil trials did not report quality-of-life results

ERG: Had concerns about plausibility of pooled end-of-treatment results being used to derive post-progression health state

Committee: Appropriate to use pooled estimates from clinical trials; utility values from CORRECT were used for NICE TA on trifluridine-tipiracil

ICER impact: uncertain_direction

Evidence gaps

no direct comparisonNo head-to-head randomised controlled trials comparing regorafenib with trifluridine-tipiracil; indirect treatment comparison required
short follow upRegorafenib trials completed in 2011 and 2013; uncertainty regarding long-term efficacy
single arm evidence onlyNo evidence for observational studies showing equal efficacy; Nakashima 2020 study showed higher OS for trifluridine-tipiracil but had high risk of bias

Commercial arrangement

simple discount pas · confidential · critical for recommendation

Special considerations

Severity modifier applied Cancer Drugs Fund eligible

Cross-references

same condition — Nivolumab with ipilimumab for previously treated metastatic colorectal cancer with high microsatellite instability or mismatch repair deficiency
comparator guidance — NICE technology appraisal guidance on trifluridine-tipiracil used for utility values and post-progression treatment costs