Synta (SNTA) reported updated results for its Hsp90 inhibitor, ganetespib, in non-small cell lung cancer (NSCLC). Synta is evaluating ganetespib in a randomized phase II trial (the GALAXY trial) where ganetespib is given in combination with standard chemotherapy. This update comes 3 months after the previous read-out, which I discussed here.
Survival signal maintained
The most encouraging finding continues to be an overall survival signal in patients whose tumors are classified as adenocarcinoma (~40% of NSCLC cases). What started as a surprising trend is becoming more robust with a larger sample size and longer follow-up. Although results are still not mature and additional 3-6 months of follow up are needed, a separation between the active and the control group is clear.
The company provided 2 analyses of overall survival. One analysis included all 172 adeno patients enrolled to date (ITT analysis), a second analysis included patients who have at least 6 months of follow up. The two graphs (see figures below) share a similar pattern, with a clear separation between the curves starting from day 100.
The signal is supported by improvement in additional endpoints: Ganetespib also led to a 50% improvement in progression-free survival (From 2.8 to 4.2 months) and a doubling of response rate (from 8% to 16%).
In both analyses, median survival for the ganetespib arm was not reached and the median survival for the control arm was identical (7.38 months). Consequently, the survival advantage, a crucial parameter in assessing a drug’s potential, cannot be extracted from the graphs. The distance between the graphs at earlier time points (e.g 60% and 70% survival) is in the 1.5-3 months range. A survival benefit of 2 months is considered approvable based on other drugs approved for 2nd line NSCLC. Taxotere and Tarceva were approved for 2nd line NSCLC based on a 2.9 and 2 months benefit, respectively. The FDA recently approved Bayer’s Stivarga based on a 1.4 months survival advantage in 3rd line colon cancer.
Benefit of ganetespib, as measured by the distance between the active and control curves is more pronounced in the patients with 6 months of follow up, where it is on the verge of statistical significance (p=0.056). This may be due to the longer follow up, as many patients in the ITT analysis have less than 3 months of follow up. The next interim analysis will prove or disprove this assumption.
Ganetespib’s emerging clinical profile looks similar to Avastin, which is approved for 1st line NSCLC in combination with chemotherapy. Avastin is predominantly used in adeno-NSCLC and leads to a survival benefit of 2 months. Importantly, there are no good biomarkers for selecting patients who are more likely to derive benefit from Avastin.
It is still unclear whether ganetespib is more active in specific subsets of adeno-NSCLC. Of the 2 predefined biomarkers for ganetespib (KRAS mutation and high LDH), so far only KRAS mutation has some sort of signal although the numbers are small.
Synta already decided to proceed to a phase III study in adeno-NSCLC patients, expected to start in Q4 2012. The trial will be similar to the phase II trial but with 500 patients (instead of 250). This is a shift from its initial strategy to target biomarker-defined populations.
Although this can initially be seen as an advantage in terms of addressable market, it might also turn into an issue in a highly competitive and segmented field such as NSCLC. Therefore, in order to gain substantial market share, ganetespib has to show a survival benefit of 2 months in the overall population or preferential activity in certain subsets.
There are ~150k cases of adeno-NSCLC in developed countries. Given that many patients do not receive 2nd line therapy and the introduction of new drugs in certain subsets, a conservative penetration of 10% looks reasonable. Assuming treatment duration of 4 months and a monthly cost of $7000, this translates to ~$400M in peak sales.
This represents the opportunity reflected by the GALAXY trial but does not take into account specific subsets within NSCLC where ganetespib could be very effective, alone, in combination with Taxotere or in combination with other drugs. Ganetespib already demonstrated good single agent activity in ALK-rearranged tumors and could also be effective in other genetic rearrangements such as ROS1 and RET fusions. Together, these subsets represent more than 10% of the adeno-NSCLC cases and the potential treatment duration could be significantly longer than 4 months.
Likelihood of success
The phase II portion of the GALAXY trial is probably one of the largest phase II studies that have been conducted in NSCLC. The large sample size (240 patients) should make final results (expected in 1H 2013) a reliable indicator of phase III. In order to assess likelihood of approval, I assume:
(i) – 50% chance for ganetespib to show a statistically significant 2-month survival benefit in the final analysis next year.
(ii) – 65% likelihood of replicating phase II results in phase III (500 patients), as the phase III portion will be very similar to the phase II.
Combined, based on these assumptions, Synta has a 32.5% chance to get ganetespib approved for 2nd line adeno-NSCLC with a survival benefit of 2 months. Using a sales multiple of 5 and peak sales of $400M, the probability adjusted valuation of Synta, based on this program alone is $650M.
This excludes any other potential markets for ganetespib, such as ALK+ lung cancer. Ganetespib is evaluated across a variety of tumor types in 14 additional clinical trials and it is likely that Hsp90 inhibitors will be used for a broad spectrum of indications.
ESMO 2012 had positive data for 2 Exelixis (EXEL) compounds: Exelixis’ wholly-owned cabozantinib (cabo) and GDC-0973, partnered with Roche. I already discussed Roche’s high hopes for the drug and the huge potential it has for Exelixis in my previous post. Here, I will focus on new data for cabo in prostate cancer.
Low-dose cabo still effective
Exelixis provided an update from a phase II trial evaluating a lower dose (40 mg) of cabo in metastatic prostate cancer. The patient population in the study was quite advanced: Of the 51 patients, 71% were 3rd line patients and 45% were treated with pain medications.
Cabo demonstrated potent anti-tumor activity with 49% and 79% of evaluable patients experiencing a “bone scan response” and soft tissue tumor regression, respectively. The objective response rate was 10% and pain improvement occurred in 18 (69%) out of the 26 relevant patients.
This trial is important for 2 main reasons:
1- In the ongoing phase III trial in the metastatic setting (with 60mg cabo), side effects might require dose reduction in order to allow prolonged treatment. It is comforting to know that the bone met activity is retained even after cutting the dose by a third.
2- Cabo is intended for use in earlier settings, especially for bone met prevention or even in 1st line combination with anti-androgens (Zytiga, Xtandi), the safety requirements are harsher. Using 40mg in those settings might create a better tolerated regimen for early stage patients.
Unique clinical profile
Overall, cabo continues to be the only agent with clear activity in 3 important domains of prostate cancer: Bone mets, soft tissue lesions and pain.
The bone mets effect is unprecedented, as no other drug has ever demonstrated bone scan resolution to this extent. It is also controversial as the effect on survival is unclear (this is currently being borne out in a phase III trial, COMET-1). At ESMO, Exelixis also presented intriguing anecdotal cases where bone scan responses were corroborated by other means of imaging such as MRI.
The effect on soft lesions shows cabo is not just a bone drug with clear “conventional” anti-tumor activity. Importantly, the low objective response rate of 10% does not necessarily represent the full extent of cabo’s activity. First, objective responses often under-represent long term benefit induced by targeted agents such as cabo. In addition, objective responses do not factor in bone met activity.
Pain is an important potential differentiator and Exelixis is running a phase III trial specifically dedicated to pain alleviation in metastatic prostate cancer (COMET–2). Although pain is a difficult endpoint to assess in single arm studies, cumulative evidence from several studies implies cabo could be the only novel agent with pain palliation in its label.
Ariad (ARIA) presented initial results for its high profile drug, AP26113. AP26113 is a dual inhibitor of ALK and EGFR, 2 validated targets for which there are active drugs in the market (Pfizer’s Xalkori and Roche’s Tarceva, respectively). Ariad is trying to position AP26113 as a “best in class” drug in both segments, similarly to the strategy with its CML drug, ponatinib.
At ESMO, Ariad presented preliminary but promising activity in ALK-mutated lung cancer. 6 (67%) out of 9 Xalkori-resistant patients had an objective response. The 3 patients who did not qualify as responders include 1 patient with substantial tumor shrinkage that was diagnosed with skin cancer and 2 patients who dropped off the study due to adverse events (1 of them unrelated to the drug). Therefore, all the patients who were evaluable for efficacy experienced substantial tumor regression.
This puts AP26113 in the same league with Novartis’ (NVS) LDK378, which is the leading 2nd– gen ALK inhibitor. Novartis’ drug demonstrated a response rate of 81% (21/26 patients) in Xalkori-resistant patients, but AP26113 appears to be better tolerated based on the limited results. Ariad will start a pivotal trial in Xalkori-resistant patients next year and could have registration enabling results in 2014.
So far, Ariad is successfully duplicating its strategy for CML in ALK+ lung cancer. Just as ponatinib appears to be the most potent BCR-Abl inhibitor, AP26113 is shaping up as a “best in class” ALK inhibitor. From a competition perspective, AP26113 is better positioned than ponatinib, which will compete with 4 highly effective approved drugs. The market leader, Gleevec, will become generic in several years, which will further intensify competition in CML. In contrast, there is only 1 approved ALK inhibitor and several other agents in clinical development. If future trials corroborate the phase I results, AP26113 has the potential to grab a meaningful share as a 2nd line ALK inhibitor.
AP26113 demonstrated activity in EGFR mutated lung cancer, although efficacy was not as strong as in ALK mutants. There was only 1 PR in 9 patients who failed at least 1 EGFR inhibitor and 2 cases of stable disease. Dose escalation continues so it remains to be seen if higher doses lead to better efficacy. One important advantage AP26113 has over Tarceva and 2nd generation EGFR inhibitors is preferential inhibition of mutated-EGFR vs. normal EGFR that is present in healthy tissues. This translates to complete lack of rash, which is commonly seen with Tarceva.
We are selling Threshold (THLD) ahead of Abraxane’s phase III results in pancreatic cancer. If Abraxane’s results are positive, it will add uncertainty and complexity to TH-302’s development in pancreatic cancer.
We are adding a third position in Exelixis following Roche’s decision to start phase III with GDC-0973. The market still ignores the transformative potential of this program for Exelixis (has co-marketing rights), but an official announcement next month could change this.
Portfolio holdings – Oct 7th 2012