So what does the market really think about Synta's (SNTA) lung cancer data? 2 weeks ago, the stock lost 33% in 1 trading session following interim results from the phase II trial for the company's lead agent, ganetespib. Since then, Synta regained most of the fall, as the market digested the data with the help of supporting analysts from Jefferies and Roth Capital.
Looking at the clinical results, it is easy to understand the market's bi-polar reaction. One the one hand, there are multiple promising efficacy signals and a good safety profile. On the other, the data set was less mature than what investors had expected.
The GALAXY trial design
Synta evaluated ganetespib in combination with chemotherapy (docetaxel) for 2nd line NSCLC (non-small cell lung cancer). The trial is a large (240 patients) randomized phase II study designed as a "signal searching" trial for identifying specific patient populations that are more responsive to the drug. The trial is big enough to detect a signal even in a subset of 20% based on progression-free survival (PFS). Based on the signal in the phase II portion, Synta plans to take ganetespib to phase III in the relevant patient populations.
Originally, there were 2 subsets in which ganetespib had already shown an efficacy signal: ALK mutants and KRAS mutants. As the drug is already being studied in dedicated trials for ALK mutated tumors, the GALAXY trial focused on other subsets, primarily KRAS mutants (based on early clinical experience) and high LDH (based on preclinical evidence and scientific rationale).
Initial signs of activity
At the interim analysis, the database included 183 patients, with an efficacy signal across three subsets:
1. Adenocarcinoma subtype (114 patients, 62% of overall population)
2. Adenocarcinoma subtype with high LDH (31 patients, 17% of overall population)
3. Adenocarcinoma subtype KRAS mutants (20 patients, 11% of overall population)
In the adeno subset, ganetespib generated a signal in PFS and overall survival. Patients who received chemo+ganetespib had a median PFS of 4.2 months vs. 2.9 months in the control arm. Response rate was also higher in the ganetespib arm (15% vs. 8%). The most striking signal was in the preliminary overall survival analysis, which showed a clear separation between the two arms, which started from day 100. This analysis should be treated with caution given the low number of events, but the trend looks very promising.
In the high LDH and KRAS mutant subsets, adding ganetespib led to an impressive signal in progression-free survival . LDH-high and KRAS mutant patients in the control arm had a median PFS of 1.4 and 1.6 months, respectively. Patients in these subsets who received chemo+ganetespib had a PFS of 4.2 months, representing a 3 and 2.6-fold increase, respectively. Responses were also more common in the combination arm for both subsets.
Are the signals real?
Subset analysis relies on several factors including sample size, numerical difference, prospective definition of subset and type of endpoint. The 3 different subset analyses presented by Synta look compelling although they vary in terms of data maturity and reliability.
The adeno subset included a large sample size (114 patients, half of whom experienced progression) but the PFS difference was modest and was not pre-defined as a co-primary endpoint. The overall survival signal in adeno patients is early but the separation of the curves looks very promising.
The extent of PFS difference in the adeno group (1.3 months, 45% improvement) might not look impressive at first glance. However, PFS often under-represents the survival benefit of targeted agents like ganetespib. Two examples in lung cancer are Tarceva and Avastin, both of which are approved based a survival advantage. Avastin led to 1.7 month (38%) improvement in PFS with Tarceva adding only 0.4 months (25% improvement).
The LDH and KRAS subsets were small but the magnitude of PFS was impressive (2.5-3 fold increase) and they were prospectively defined. One encouraging sign is the fact that effect size was maintained as the data matured during the past 3 months, according to Synta's CEO. Although the KRAS subset is the smallest, it is the only one relying on a molecular characteristic of the tumor with a clear rationale and clinical activity for ganetespib in an earlier trial as monotherapy.
No statistical measures were provided by the company, due to the early stage of the data set. Nevertheless, Synta's decision to proceed to phase III implies that there are at least statistical trends.
Synta vs. Array
In KRAS mutated lung cancer, the most relevant comparator for ganetespib is AstraZeneca's (AZN) selumetinib, which I wrote about in my ASCO 2012 summary. Originally licensed from Array Biopharma (ARRY), selumetinib was also added to docetaxel for NSCLC patients with KRAS mutants.
An indirect comparison between the two agents shows a similar PFS benefit over docetaxel, with a ~2.5-fold improvement. Response rate for selumetinib appears higher, but it is unclear whether responses were defined in the same way (confirmed vs. unconfirmed).
Selumetinib's data set is larger and more mature, automatically making it more reliable. The difference in PFS and response rate were also statistically significant with selumetinib, whereas there was simply not enough data to evaluate statistical significance for ganetespib. Lastly, the selumetinib trial was a double blind study in contrast to the ganetespib trial, where physicians knew which patients were receiving the drug or placebo. This could have created an investigator bias in evaluating PFS and response rate. Needless to say, updated results for ganetespib could change this comparison to either direction.
Both drugs are expected to enter phase III in KRAS mutants later this year. Since they have distinct modes of action, they could be synergistic in combination. It is unlikely that evaluating the two drugs in combination will occur before they reach the market, as they are running neck and neck towards approval. The only company with an Hsp90 inhibitor and a MEK inhibitor in clinical testing is Novartis (NVS).
The PFS benefit across the 3 different subsets looks reliable thanks to the large sample size for the adeno subgroup and the fact that KRAS and LDH sub-analysis was prospectively defined by the company. The biggest surprise was the preliminary survival trend in the adeno group, which represents a large commercial opportunity ($1.5B in the US alone). Even if only one of the smaller subgroup analysis is corroborated by larger studies (LDH or KRAS), ganetespib is still looking at a $0.5B opportunity, excluding the ALK opportunity.
Still, one cannot help wondering if Synta should have waited 2-3 more months to provide the investment community with the complete picture. The next update is expected at the ESMO meeting (starts Sep 28th), which is expected to include more patients and longer follow up.
We are initiating a position in Genmab (GEN.CO) following positive data with its CD38 antibody, daratumumab (discussed here). We are also selling our Onyx (ONXX) position at a profit of 84%, as high chances of accelerated approval for carfilzomib are already factored in.