LOXO, BPMC and genetically determined cancers.
The conversation on these new TKIs often dances around the difficulties of finding treatable patients so I thought I'd run some numbers. The purpose of running these numbers is to set up a framework for conversation:
LOXO believes that there are 1500 to 5000 new cancer patients a year in the US with NTRK fusions. Call it 3000. And, so far, they have included in their trial 17 different cancer types - some of which are quite large indications. All told the total number of people with cancers of the 17 types (including lung, colon and breast) they have included so far are probably over 500,000 annually in the US. Call it 600,000. So they are predicting 0.5% will test positive for the fusion.
That means that in order to find a single patient they have to give a test to 199 people to find 1 person for whom their treatment will work. And note that they explicitly note that the general tests aren't always reliable. If the test costs $1000 (which is probably a low estimate if their test is a unique test, instead of being part of a single panel for multiple mutations) then that is spending $200k to find a single patient.
By way of contrast, BPMC is talking of finding about 5% of patients with PDGFR mutations - and using the same test costs that translates into 'only' $20k to find a patient.
Note that LOXO, in the past, has talked about attempting to enrich the patient population by finding histologic or origin types that have much higher rates of the fusion - e.g. "7 of 28 younger patients (25%) tested TRK fusion positive [in papillary thyroid cancer]". Obviously this is hugely helpful if it bears out - but this isn't emphasized, and isn't mentioned at all in many of their more recent releases. (Note - following the enrichment strategy significantly shrinks their patient population since, for instance, the indication above is around 1000 patients per year total in the US. Leading to 250 patients treatable by their TKI.). My guess is more than 1/2 of the treatable patients in their estimates are in the most common cancer types at extremely low incidences (e.g. 0.3%) and it isn't clear they will ever be addressable until there is a common panel given routinely like histologies are done today.
Finally, note that both BPMC and LOXO are pursuing RET inhibitors in lung cancer with an incidence rate of around 1 to 2 percent. Call it 1.5%. I'd consider this on the edge of addressable - but would become much more so if it could be combined in a panel with an ALK test.
AI
