Last month’s failure of Infinity’s (INFI) Hsp90 inhibitor (retaspimycin) has mixed implications for Synta (SNTA), whose Hsp90 inhibitor (ganetespib) is in phase III for lung cancer. On the one hand, ganetespib’s most advanced competitor is now eliminated but on the other, the failure raises doubt about targeting Hsp90 in lung cancer. In that sense, Infinity’s failure gives another reason to question the validity of Synta’s phase II results (that are already quite controversial). As background, I discussed Synta’s results here.
Are all Hsp90 inhibitors interchangeable?
It is still unclear whether retaspimycin failed because Hsp90 inhibition is ineffective in lung cancer or because the drug itself is not a good Hsp90 inhibitor. Infinity’s drug belongs to the first generation of Hsp90 inhibitors that are considered less potent and more toxic than second generation drugs such as Synta’s ganetespib and Novartis’ (NVS) AUY922. Therefore, any failure for retaspimycin could be explained by the fact that it does not lead to the deep and broad Hsp90 inhibition required for clinical benefit. For example, one can claim retaspimycin-associated toxicities and poor PK profile prevented Infinity from reaching effective exposure.
Retaspimycin is probably an inferior drug but it is clearly active to some extent, at least in ALK-mutated lung cancer where the drug demonstrated activity in 3 patients. This appears to be an Hsp90 class effect, validated by responses with ganetespib and AUY922.
Infinity’s phase II trial evaluated retaspimycin in a similar (but not identical) setting to Synta’s phase II trial: 2nd/3rd line NSCLC in combination with Taxotere with overall survival as the primary endpoint. One major difference is Synta’s focus on adenocarcinoma vs. Infinity’s focus on smokers and squamous carcinoma. Both trials were large enough to be able to generate a survival signal.
Infinity published only top line results, which makes it challenging to draw a final conclusion, but the press release gives the impression there were no signals across the different subgroups (including adeno tumors).
Same target, different outcome
Are drugs that employ the same mechanism interchangeable? Not always.
In oncology, there are many examples of 2 drugs that looked similar “on paper” but greatly differed in terms of clinical activity. This discrepancy cannot always be explained, although in some cases there are hypothetical reasons ranging from biochemical features to dosing and clinical design. Below are 2 historical and 2 recent examples.
Example #1 – Tarceva and Iressa – Both drugs are EGFR inhibitors with a similar activity spectrum in cells and animals. While Tarceva led to a survival difference in its phase III, Iressa failed to do so and was consequently withdrawn from the market. Interestingly, both drugs have strong activity in EGFR-mutated lung cancer but only Tarceva had activity in the entire patient population. The underlying cause for the difference is unclear. Some suggested that Tarceva led to better target inhibition as it was dosed at its maximum tolerated dose in contrast to Iressa, which was dosed at a third of its MTD.
Example #2 – Yervoy and tremelimumab – The 2 anti-CTLA-4 antibodies were developed in parallel for melanoma. While both agents demonstrated activity in the form of exceptionally durable responses in <10% of patients, only Yervoy led to a survival difference in the entire patient population in phase III. It is unclear whether tremelimumab is truly an inferior agent, with some blaming dosing regimen and inclusion criteria as potential reasons for failure.
Example #3 – Ibrutinib and CC-292 – Both drugs are covalent Btk inhibitors with activity in CLL, among other blood cancers. Ibrutinib is much more advanced and follow up for CC-292 is limited, but based on available data ibrutinib looks significantly more efficacious in CLL. The reason for this difference is still unclear given CC-292’s potency and selectivity profile.
Example #4 – CO-1686 and AP26113 – Although the two drugs have a different profile (CO-1686 is a covalent inhibitor selective for mutated forms of EGFR, AP26113 is a dual ALK/EGFR inhibitor), both were shown to inhibit cancer cells with a “Tarceva- resistant” mutation (T790M). Despite strong preclinical results in EGFR mutated cancer, AP26113 failed to generate activity in patients with T790M lung cancer. In contrast, CO-1686 had clear activity in T790M patients with responses in 3 out of 4 patients treated at the highest dose tested.
Exelixis’ MEK inhibitor as a fresh example
Genentech recently published a research paper which demonstrates how 2 different inhibitors of the same target (MEK) can have different activity profiles, even at the molecular level.
Researchers compared 2 MEK inhibitors: GDC-0973 which was licensed from Exelixis (EXEL) and GDC-0623 which was developed internally. Exelixis’ MEK inhibitor appeared more effective in tumors with BRAF mutations whereas the internally-developed MEK inhibitor was more effective in KRAS-mutated tumors. This difference was explained by the distinct MEK activation states each mutation confers (MEK is downstream of BRAF and KRAS).
The implications for Exelixis are mixed, as results bode well for success in BRAF-mutated melanoma but also imply the drug is not suitable for KRAS mutated tumors (especially lung cancer), which represent a larger commercial opportunity.
It is important to note that Synta still has a large data set with a real signal that can justify the ongoing phase III trial and the inclusion criteria. In addition, as Hsp90 is involved in the function of so many client proteins, it is plausible that ganetespib has a different spectrum of activity resulting in a better inhibition profile. Lastly, Hsp90 inhibition has great potential in molecularly defined subsets of lung cancer (ALK, ROS1 and RET rearrangements), not to mention additional indications.
While I have strong conviction in ganetespib’s superiority over other Hsp90 inhibitors, the approach of Hsp90 inhibition in unselected lung cancer patients is still unvalidated. For now, Infinity’s failure slightly decreases the attractiveness of Synta’s approach.
As a result, we are decreasing exposure to Synta by selling one of our 3 positions. We plan to hold the current position going into data readouts next year.
Portfolio holdings – October 6th, 2013