The difference between entrectinib and larotrectinib as well as most other oncology drugs is that the former are prescribed based on a genetic marker of the tumour (i.e. NTRK fusion-positive), regardless of the tumour’s location, while for the latter prescription tends to be largely based on tumour location. Consequently, tumours in many different locations may be treated with entrectinib/larotrectinib, including – but not limited to – tumours in the lung, colon, pancreas, breast, thyroid gland and salivary gland.

Given that the two TRK inhibitors can be used to treat tumours in many different locations, the total number of comparators equals the sum of the number of comparators for each tumour site. Included comparators are, among others, bevacizumab, nivolumab, sunitinib, eribulin, docetaxel, regorafenib, pazopanib, and pembrolizumab. The structure of the model consists of a decision-tree, reflecting the testing phase, combined with a partitioned survival model (Roche global model). Based on the modelling guidance recently developed in the HEcoPerMed project, focus areas in further developing the model are: adequately capturing expected NTRK testing practice; ensuring that test properties are accounted for; addressing any mortality during the testing phase; exploring the estimation of the (as of yet unknown) prognostic value of NTRK-fusions; exploring any relevant ‘values beyond the QALY’; and integrating structural uncertainty in the uncertainty analysis.