Partitioned survival model

Entrectinib, 100 mg capsule, $47.67

• breast cancer (secretory): paclitaxel, docetaxel, carboplatin, eribulin
• breast cancer (non-secretory): paclitaxel, docetaxel, carboplatin, eribulin
• colorectal cancer: pembrolizumab (MSI-H), FOLFOX, FOLFIRI, bevacizumab + FOLFOX
• MASC: sunitinib, gefitinib, cisplatin + gemcitabine
• lung cancer (squamous) (SCLC): pembrolizumab + carboplatin and paclitaxel/nab-paclitaxel, pembrolizumab, docetaxel
• NSCLC: pembrolizumab + pemetrexed and cisplatin, pembrolizumab, pemetrexed + cisplatin, pemetrexed, cisplatin, docetaxel
• neuroendocrine: octreotide
• pancreatic: FOLFIRINOX, gemcitabine + nab-paclitaxel
• soft tissue sarcoma: doxorubicin, imatinib, eribulin
• thyroid cancer (papillary): lenvatinib, sorafenib
• thyroid cancer (other): doxorubicin, paclitaxel

Naive comparison based on literature estimates for comparator PFS and OS; 1 trial selected per comparator. Comparator populations were not selected for NTRK fusion status

• Pooled analysis masks the variability in the comparative effectiveness and cost-effectiveness of entrectinib across tumour sites. This in turn masks the patient populations, settings, or conditions for which entrectinib may or may not be cost-effective.
• Pooled analysis does not represent the heterogeneity in response, duration of response, PFS, or OS as reported in the clinical report. Sponsor’s analysis relied on survival analysis (estimation of PFS and OS curves) performed on a highly heterogeneous population in terms of prognosis, based on tumour site and number of prior lines of therapy, which is inconsistent with the core assumptions of survival analysis (requiring a homogeneous study population). The averaging of outcomes across comparators that vary in costs of treatment and prognoses was performed incorrectly because it did not account for the changing composition of the population over time.
• Stratified analysis is presented for 6 cancer subtypes, whereas reimbursement is sought for at least 17 adult cancer indications (cancer types, including subtypes, represented in the clinical trial data) and potentially more cancer subtypes in which NTRK fusion mutation is present, but for which there is no clinical or health economic evidence. For all stratified analyses presented, the sponsor assumed the PFS and OS of the entrectinib arm were the same, regardless of tumour site, without clinical justification and in contradiction to the heterogeneity reported in the response rate and duration of response outlined within the CADTH clinical report.
• The costs of identifying patients with NTRK fusion mutations are underestimated. The sponsor assumed that patients would largely be identified using IHC; however, CADTH clinical experts described IHC for detection of NTRK fusion mutations as still under development and not clinically validated with a known test accuracy for all tumour types. Clinical experts also indicated a strong preference for NGS testing because it can screen for multiple mutations at once without destroying the patient’s pathology sample.
• The sponsor’s analysis extrapolated PFS and OS survival curves without assuming any treatment waning and, as such, under-representing the uncertainty of predicted long-term outcomes substantially past the observation period for specific tumour types.
• The sponsor excluded any subsequent therapy costs for those who fail on a first-line therapy. If entrectinib was to be used in a first-line setting, the treatments it replaces would likely be used if the patient were to progress on entrectinib. The sponsor also excluded relevant health care costs that would be incurred by the patient over their lifetime.
• CADTH identified numerous errors in the sponsor’s model, such as using the height of an individual to determine dose rather than weight.

• The CADTH reanalysis included the following changes: corrected costing errors; used tumour-specific PFS and OS data; applied different extrapolation methods to OS and PFS; applied relevant testing costs; considered a greater number of relevant tumour types; presented results for each tumour type; and included a scenario analysis comparing entrectinib to first-line therapies with and without subsequent therapy costs.
• For the pooled analysis for second-line therapies:
  • The ICER of entrectinib compared with BSC, averaged across all tumour sites, in patients known to have NTRK fusion cancers is $1,272,991 per QALY gained.
  • No level of price reduction for entrectinib will achieve an ICER of $50,000 per QALY on the average cost-effectiveness across all indications.
  • Incorporating the costs of case finding using NGS testing, the ICER of entrectinib compared with BSC care increases to $16,746,589 per QALY gained. There is no price reduction for entrectinib that will achieve an ICER of $50,000 per QALY.
• CADTH notes there is a substantial amount of heterogeneity in the cost-effectiveness of entrectinib across individual tumour sites, with the ICER varying from $94,645 per QALY gained for MASC compared with sunitinib to higher costs but fewer QALYs for entrectinib compared with relevant comparators in various tumour sites (thyroid and CRC).
• As a scenario analysis, CADTH analyzed the pooled analysis for comparators the sponsor identified as first-line therapies:
  • The ICER of entrectinib compared with BSC, averaged across all tumour sites, in patients known to have NTRK fusion cancers is $2,057,174 per QALY gained.
  • A price reduction of 82% for entrectinib may achieve an ICER of $50,000 per QALY on the average cost-effectiveness across all indications. However, this assumes patients on entrectinib would receive no subsequent lines of therapies. If subsequent therapy costs are incorporated, there may be no price reduction for entrectinib that will achieve an ICER of $50,000 per QALY.
  • Incorporating the costs of case finding using NGS testing, the ICER of entrectinib compared with BSC care increases to $9,209,215 per QALY gained. There is no price reduction for entrectinib that will achieve an ICER of $50,000 per QALY.